Shb | Kami Vaniea

Security and Human Behavior 2026 Day 2

Tue, 19 May 2026 12:18:48 -0400

Welcome to the 19th Security and Human Behavior. The write-up below is a live-blog of the workshop.

Contents:

- Speakers: Carolin Lämmle, Ryan Wright, David Reeves, Samantha Phillips, Tony Vance, Laura Arno
- Speakers: Joseph Bonneau, Kami Vaniea, Serge Egelman, Jayati Dev, Michele Massberg, Ryan Shandler
- Speakers: David Sidi, Avinash Collis, Geoff Tomaino, Andrew Odlyzko, Tawfiq Alashoor, Blase Ur
- Speakers: Sascha Romanosky, Susan Landau, Jean Camp, Matt Blaze, Alessandro Acquisti, Jeremy Epstein

Also see:

2026 ,
2025 ,
2024 ,

Session 5: Organizational Security

Speakers: Carolin Lämmle, Ryan Wright, David Reeves, Samantha Phillips, Tony Vance, Laura Arno

Carolin Lämmle

Title: System Administrators - Power, Gender, and Ethics

Systemadmins - interviews with 11 male administrators then later a study of 13 female system administrators.

Unsuprisingly female system adinistrators are not common on teams leading to many issues

Surveyed a larger set of system administrators (32 Women, 216 men, 7 other)

I am in a powerful position <- both genders agree they have power, are irreplaceable, respected by suepervisors, relied on for advice, and comfortable with with giving orders. They also were self-aware that they could do quite a bit of dammage.
Struggles with doing things “quick and dirty or the right way”
Honor code seen as important
Gender differences:
- Men more likely to say “I have skills no one else have”
- Women less likely to agree that they have special skills, are competent

Ryan Wright

Title: Does compliance Actually Reduce Cyber Incidents: Two decades of acadiemic research. Seven CISOs.

How much did your organization spend on compliance training, awareness, and policy enforcement last year? Evidence?

Does compliance actually matter? Maybe it is the wrong construct.

Pre-registered study. Which is challenging to do.

When talking to CISOs about what is working and what is not working for them. Issues like influence, trust, shared goals, regular communication and genuine partnership. AKA .

Findings

Compliance didn’t move the needle

Compliance does matter for smaller organizations, so having compliance does matter, but adding more maybe doesn’t

Shared social capital is big

If both cyber and business and social capital are high and balance - then less odds of a severe breach

Who benifits the most

Big organizations need more social capital
Top management doesn’t understand cybersecurity plan? It will probably fail.
Security teams that act as community members are more effective

“Security leadership increasingly resembles community stewardship”

Compliance is important to get security up to a certain level, but past that there are limited returns from just compliance.

David Reeves (Virginia Tech)

Both a PhD student and a CISO

Behavioral distortions in organizational cybersecurity decision-making

Impact of anticipatory feelings on organizations cybersecurity decision-making

What level of risk is acceptable? Having these conversations at the board-level is complex.

Large data breach
State said that acceptable risk was at a certain level, but the on the ground decisions did not match that high level goals
Money saving was an aspect
Costs of breach were serious and not necessarily properly thought about before hand
- Loss of trust with partners - long term loss
- Reassigned employees from other projects to handle the situation
- High levels of burnout
- Admins took lots of blame, even though the risk decisions were taken by others
- Breach costs like having to pay for LiveLock for people who’s data was lost.

Core argument: Organizational leaders rely on cognitive biased interpretations that systematically distort cybersecurity risk perception.

Mixed method study: quantitative data such as financial, ceybrersecurity assessment results, incident informaiton, organizational environmental data. Qualitative data like grounded theory interviews of populations like CEOs, CFOs, CIO, and CISOs.

Samantha Phillips (The University of Tulsa)

Imagine two organizations A and B and the same security intervention is used at both, but the outcome is quite different. What about security culture is causing such a difference in how an intervention is recieved?

Used:

Hofstead’s Organizational Culture Dimensions
Schein’s Three-Level Model of Culture

Research study with survey study and follow-up interviews.

Employees did think that the company cared about security, but they were unsure if they would actually get any help if something happened
Leadership story
- Operations leader, deployed security cameras in a site, then security team found out. Security team didn’t approve, and opperations had to remove the cameras. Leading to frictions and tensions in the culture.
Another story
- Security team depoloyed MFA, so they sent it out to the rest of the company
- But…. many employees did not have phones, lots of worked sites had no cell service. So MFA caused bad security behaviors like everyone using the one account they could get logged in.

Take aways

Measuring the type of information security culture is important to target interventions
- Easy going culture may not match well with a strict security intervention
Assess culture from multiple dimensions/perspectives
Consider measuring pre/post to understand how an intervention is impacting a culture

Tony Vance (Virginia Tech)

Title: Rethinking security culture

Story: including gaining access to a private key - which is very not good

Enbled attackers to access email addresses of many organizations that use Microsoft email products
Review was done and publicly released
- Found that security culture needs to be improved at Microsoft
- But what is “securiyt culture”? Report did not define the term.

Risk management

Measuring scales for culture

Security education and awareness - good overlap among scales
Many other constructs measure a range of things. 83 different constructs found in the review

Some constructs are correlates, some are determinants, and some outcomes. These are different aspects entirely.

Design of survey scales are not at the level of modern scale development. Issues like properly validating the scale to make sure it measures what the authors intended. It is important that participants answer questions in a way that matches the researcher understanding, if this is not tested then there may be a big issue.

A scale development cycle:

Conceptual definition
Dimensioality specification
Item generation
Content validity

Wants to improve how culture is measured so organizations can not only say they want to improve, but measure what their current state looks like.

Laura Arno

Talk: How security Policy Motivates Computer Abuse: Organizatoinal Technology Injustice

Overly tight security policies can lead to shadow IT, where users go around the security and setup their own IT which then results in overall lower security.

Insider threats

Mone going towards employee monitoring and surveillance as a way of mitigating threats.

Tensions between the IT and the non-IT employees. This an cause computer abuse.

Lots of research focuses on compliance as an outcome. But recent research is calling this into question. Measuring non-compliance is also becoming something to measure. But non-compliance is often conflated with negligence, cyberloafing, violation and computer abuse. But much of non-compliance is closer to work-arounds of overly strict security policies that make day-to-day work very challenging. We should consider measuring non-compliance as a metric of how badly a policy mismatches with employee needs and willingness.

Many employees are not aware of the policies. But they know what they think the policies are. They may only know that the policy is restricting them, that they have lost control, and that they do not like them.

Perceptions of the restrictons is driving their behavior.

Q&A

If we talk about cognitive biases, are we suffering from them? Is phishing really the biggest risk? CISOs are doing training/compliance because NIST has said so. Are we the biased ones? Should be re-assessing good protections?
- Yes: having academic conversations where we have crazy conversations and then testing them is really great. It causes a slow building of understanding that is very valuable.
- It is hard to measure security culture, especially if it is a suplier that was hacked
- We spend all this time and energy on barriers. But we need better partnership with employees. Reporting and response are also super important. It isn’t all up-front protection.
For Laura - you mentioned shadow IT and that the policies are unjust
- Studying more than compliance is important. Shadow IT is very much a thing. It is possible to both be compliant (long password changed often) and not in compliance (password written on white board because no one can memorize it).
Social capital
- Ryan: in our study we had the two roles measure each other’s social capital. It was important to see not just how people see themselves, but also how other see them.
It is not that compliance does not work, the issue is that compliance is not done correctly. Box ticking vs finding underlying problem and addressing it. In lawsuits: a security report is often based on a CISO answering a questionare, not based on what was actually going on.
- We need to re-think what compliance actually is.
Relational governance - social capital. It was interesting in our case we saw that you cannot order people like tenured professors or members of parlement. So social capital is vital to get security actually done. How do CISOs navigate the changing roles
- Ryan: CISOs: they feel they need to empower users, empower business units, and they need a shared risk model
- David: It is imporant as a CISO is to weirdly not be a technologist. But in a CISO role the position is to be a collaborative leader, not a technologist. And to figure out how risks are applied to the company’s business function. Role is to think about what the organization risks are not the IT risks.
To Ryan: In political economy people focus quite a bit on trust because it can be measured and put in a model. Would it help to start with trust in these measures of social capital and collaboration within an organization.
When I think about security culture I think about “I know it when I see it”. How much cluture is based on the technology of the moment and how much is actually culture.
- Hopefully not the technology, because that is always changing. Culture also changes over time. So both are combined.
- They are quite independent. Even if a tech stack was swapped out the culture should persist, though it would need to adapt.
- At home I’m productive (no restrictions) at work I’m quite restricted so I’m less productive.

Session 6: Trust and Security Public Policy

Speakers: Joseph Bonneau, Kami Vaniea, Serge Egelman, Jayati Dev, Michele Massberg, Ryan Shandler

Notes in this session were done by my incredible student due to me being in the session.

Joseph Bonneau

Title: Cryptographically Verifiable Lotteries

Making lotteries verifiable for cryptography Lotteries are used for many purposes like moose hunt lottery for hunting permits, judge assignments, conscription, and security screening Case Studies: biased randomness US conscription lottery, 1969: those with later birth months were not selected Diversity visa program, 2011: results are void as it was biased

How can we achieve a verifiable lottery? Physical randomness ceremonies: the issue with this is that physical randomness can be faked: hot and cold balls used to aid cheats Randomness from natural phenomena or stock market from asset prices Levels of Verifiability: Using cryptography we can use multiparty randomness beacon protocol (i.e. League of Entropy) the trust model here is that if a majority of them is good the randomness will work well

Participatory protocols: Only 1 of n correct participants needed for security but no security downside to adding participants

Kami Vaniea

Many conferences this year are struggling with high submission rates and potentially AI generated content

What is SOUPS: Symposium on Usable Privacy and Security SOUPS is focused on human factors of security and privacy technologies and have started in 2005. Kami was technical paper chair for 2025 and 2026.

There has been a boom of paper submissions:

2024 was 156 submissions with 21% accepted
2025 was 157 submissions with 19% accepted
2026 was 251 submissions with 15.5% accepted

251 submissions and 15.5% (39 accepted submissions( and 3 main reasons for number of submissions)

AI may be a culprit here - its a common source of blame this year
CORE ranking- SOUPS was just raised to A-level in late 2025 (being A-level is a big deal)
Location being in Germany

Statistics: Kami asked reviewers to let us know about problematic AI and some issues identified included mangled references, fabricated references, citing text does not agree with cited paper, or poorly written text or text that does not match the rest of the paper

18 flagged by reviewers as possibly having problematic AI
12 rejected in round 1
3 rejected in round 2
3 accepted both were flagged for having potentially AI generated text Problem: we never looked this closely at references before

Step 11: SOUPS will be using student volunteers to review the output of reference checkers for all accepted papers!

Serge Egelman

Title: Is it time for software to put on its big boy pants?

Most engineering depends on complex supply chains: Apple doesn’t make glass, Toyota doesn’t make brake pads Airlines use chicken cannons to test airline parts. For engineering software, you should be using ⅙ of the time coding and most of the time in planning and validation. Products must list ingredients and/or share any hazardous ingredients but this doesn’t exist in software. Part of the problem is that “developers usually choose which corporates they incorporate and compliance / QA folks may be left in the dark”.

Some developer SDKs take location Data and there is a paper on this

Awareness of app behaviours and some developers just did not know Data sent to measurelib.com and it was linked to a defense contractor and there is a book by Bryan Tau on this topic Physical products that cause harm when used as intended are subject to recalls yet we don’t have this for software

Methods for accountability exist: civil engineers are licensed Where do we go from here?

SBOMs
ONCD organized a working group to come up with a framework for software liability
Maybe engineers need to be professionally licensed

Jayati Dev

Title: Multistakeholder approach to open source policy

Open source software is everywhere and for maintainers the bigger is gets the harder it is to maintain and for users there is usage in insecure context and hard to get support

Can we add policy “rocks” for modern infrastructure/ open source software as many policy conversations happen in silos. What would a multi stakeholder policy approach look like for open source? At cybersecurity policy workshops they had breakout sessions with academic, industry, and government 20-25 participants

Initial Findings: a case study to understand the 3 way communication participation via a conservatorship model to fund critical open source, joint bug bounty programs, and support secure implementation of open source. Liabilities and how do we assign liability and who do we assign liability protections for researchers. Having IP and safe harbor protections for open source developers

Plan to convert this to a formal study, conduct more workshops, and collaborate on evidence based research

Michele Massberg

Title: Decrypting Covert Operations: Human Factors in Cryptographic Design

This study presents the first known decryption of Covert Operations panels and the problem:

In order to decrypt this: they transcribed the sculpture to string representation in Python 3.12.0 And then ran statistical analysis and computed annotation using leiden conventional sigla (Dow 1969). If a certain symbol was missing, they would use a different symbol.

Looking at these statistics, the arabic panel looked like it was plain text and english, cyrillic, and 3rd language looked like something else. After a bruteforce comparison, she used a ciphertext only key search and was able to decrypt this text and reconstructed and cracked it again.

For the 4th panel, a reconstruction of the plain text using key and line number. Through reconstruction of cyrillic panel, it was an executive order during the cold war.

This work provides the first complete technical resolution of the four Covert Operation panels. resolution of the four Covert Operations panels. Across both the English and Cyrillic reliefs, the ciphertexts were found to be consistent with periodic polyalphabetic substitutions, implemented as keyed Vigenère variants. The cipher design reflects the practical constraints of artistic design, manual inscription and a preference for mnemonic, thematically aligned keys.

Ryan Shandler

Title: Destabilizing Democracy: The Long term societal effects of cyber operations

When we think of the impact of cyber attacks and threats, computer scientists think of denial of service, degraded infra, data theft but Ryn thinks about societal divisions, distrust in government institutions, elevated perception of threat, and support for anti-democratic policies.

Phase 1: Attacks arouse psychological distress (increased anxiety levels) For individuals who are not technologically adept, they express significant anxiety

Phase 2: Emotional reaction triggers behavioural/political shifts Cyberattacks even when they don’t cause a lot of damage, for political feelings cause signficiant damage and impact

Weaponizing this phenomenon? DARPA shared that foreign entities could use this research to launch attacks against the US. This may be similar to a termite attack.

Behavioural Model of long term societal effects triggered by adversarial cyberattacks and there is always an effect that happen immediately afterwards.

The experiment that starts tomorrow will have 5000 participants experience the treatment/control and record short term, medium term, and long term effects over the next 5 years. Some attack source will be China or unknown source or damage to mobile networks (shutting down internet access) or US response = partisian divisions. This will be consumed via social media feeds and long term news sources. For individual effects it’s usually seen for 1-2 months but will see a long term observation of societal effects and multi country analysis. How do we mitigate democratic destabilization.

Results will be shared at SHB 2027!

Q&A:

Ryan was asked, “Do you think there will be something similar to the Baader-Meinhof phenomenon for cyberattacks: If you buy a car, you see the car continuously” there isn’t just one cyberattack it’s whats the effect of one cyberattack followup in your daily life and maximizing external control and impacts in the long term

Question for Kami, “NSF has banned from using AI to assess anything about them. What would you do for 50,000 proposals that are to decide to spend $0-50 billion dollars?” If it’s purely AI generated, it’s a lot easier to check them and be cruel in the punishment. The discussion sections suffered far more from AI and they used AI for the sections where it’s the hard part where you have to think about it. We need policies on this. There was discussion on requiring a DOI for each citation for paper submissions and avoid having to revoke any papers due to a hallucinated reference.

Ryan got asked, “With so many cyberattacks already in the news, why will this particular treatment have a great impact?”. Ryan says his expectation is that they are trying to create a larger scale of cyberattacks and be realistic that it will break through the media cycle. They are taking a real attack and amplifying it a bit to see if their anxiety will be amplified so participants can look it up later (and be asked about this).

Kami was asked, “One of the references was not real when they submitted a paper and when they used AI to edit the paper it changed the name of the paper.” Kami said this could be a good learning opportunity for students and researchers.

Kami was asked, “Should this be handled by a higher level?” Kami said it hasn’t bubbled up yet and everyone is trying to handle it this year frantically. Another question, “I’ve had an issue with AI generated reviews, is this an issue with SOUPs?” Kami said some reviewers will not be asked back and only 1 reviewer she had a suspicion used AI due to lack of content and constructive feedback. “Across Security and Privacy papers, the people who are writing the Discussions sections just don’t seem as developed?” Kami got another question, “At the end of the day, we shouldn’t hate those who use AI but at the end of the day we should care if a paper is using AI but care about the quality and be smart about evaluating quality”

Session 7: Public Policy and Privacy

Speakers: David Sidi, Avinash Collis, Geoff Tomaino, Andrew Odlyzko, Tawfiq Alashoor, Blase Ur

David Sidi

Access, Privacy, and Conviality with We Build Networks

Tech for advancing values > Broader Access to powerful tech > institutional support > participipator > enthusiastic

Working on making networking and networking tools more accessable to people. Workshops on topics like ToR node setup.

“We build networks” working with public groups to do educations.

Avinash Collis

Ads

Positive for users - informational role, match buyers and sellers
Negative - higher prices, hyper focused

Facebook’s internal A/B testing platform has a small number of users who never see ads. They do this for experiment purposes. They recruit people from the ads and no ads. Do an incentivisation study by paying them to not use Facebook. In theory this will help determine how valuable ads are financially.

“Would you be willing to stop using Facebook for one month for $40?”

No significant differences between ads and no ads group. Both gropus value Facebook at about $31/month.

In the next project: how does this generalize across othe internet? How about other platforms. Also targeted vs non-targeted ads. This study looked at willingness to keep being in their condition. This experiment is mid-way.

Users in the ad blocking condition (where people have ad blockers) group want higher payments to keep using the system - this means that they are less willing to keep using the advertising blockers than put up with ads
Those who experience ad-blocking

Geoff Tomaino

Title: The role of preference ordering in consumer privacy violation perceptions

The value consumers get out of advertisement

First-order preference: What do you want?

When you went out for lunch, you wanted a sandwich. Second-order preference: What you want to want
They want to want things that they don’t actually want

People take the view that advertisements represent what the brand thinks my first-order preferences are?

For topics like news, food, people often want to want something other than what they really want.

Study 1:

We asked to imagne thy used an investing site named Finance
FinanceWise recommendation: Low-risk investments
Gave other recommendations that match first or second order preferences
Found mismatch condition makes people feel like a privacy invasion

Study 2:

Your recommendation for
First order: indulgent foods
Second order preference Healthy OR indulgent foods
How do people feel about
More of a violation when ad goes against second order preferences

Andrew Odlyzko

Title: Where is our society and economy going?

The modern erra might be named after Turing from Turing Pharmaceuticals who managed to massively marked up the drug. Amazingly the drug was a generic drug.

Government has sorta given up on anti-trust enforcement - not just current administration
Increasing volume of information - information in the eoconomy as well as private information
- Cartels
- Growing part of the economy is corperate profits

Lots of this has to do with the great enrichment of communities. The industrial revolution.

The market as you think about Adam Smith’s conception. market depends on a certain amount of capacity. Buyers and sellers come together where it is all about quantity and price.
But now with more information available. Now the incentive is to get as much info about your buyers as possible.

Price descrimination is becoming more of a thing. Most of it is done is hidden forms. But this is very unpopular with the public. So most of it is hidden.

Get credit card data on potential employees so they can decide who is desperate enough for a low paying job.

Tawfiq Alashoor

Title: Securing Digital Transformation in the Age of AI: Behavioral Privacy Penetration Testing

Long ago no one understood the value of salt. Then it became rare, and became a traded form of money. Solariam (salt) causing “salary”

I’ve heard that data is the new money. The value of personal data is raising like enver before.

We do allot of security penetration testing, but we don’t think about ethical ways to test the human brain to find vulnerabilities.

In cybersecurity we want to avoid a single point of failure. But our testing is focused on the technology, the firewalls. This single point of failue manifests through bad privacy decisions.

We asked 2k people a set of questions:

Subjects that were not primed to privacy and not nudged to privacy answer less questions than those who were primed and nudged
Even with small sample sizes we see the same thing

When a robot violates privacy who is to blame

Consumers blame robot
Companies blame other companies

AI arms race

contact with AI is social media (attention)
contact with AI is GenAI (intimacy)

Privacy is Dead <- today’s myth

Blase Ur (University of Chicago)

Title: What could data subject access rights be?

Built a Tracking Transparency Tool

US FTC’s FIPP’s

Notice - historical focus
Choice - historical focus
Access - interesting future focus by us
Integrity
Enforcement

When you download your Twitter data there is this targeting.js file explaining targeting ads. More data here than in the public user interface on why this ad.

Facinating targeting stats. Including things like race.
This is sorta public data, why are people not looking at it?

When you download data you get a huge zip file or a huge file. How do you parse.

Some files were millions of chars but no line breaks
Very hard to parse, hard to read
Unix timestamps simple example
There are no definitions for what some of the things be in the files.

how can we make some of this data more visible and actionable
Have users annotate the data and share with researchers
They wondered
- What data is stored
- How is it stored
- How is it used
- Takeaways from the data
Necessary features for a useful tool:
- Definitions and explanations in context
- Searching and filtering
- Data deletion and modification in context

Q&A

How does first order or second order align with actual vs ideal self. You are being targeted based on your actual self. Is it because I have a “shady” preference. And that causes the privacy violation.
- Embarassement is coming from inability to connect first and second order preferences. It suggest a lack of control over self.
Is it more about the ads or about being targeted?
- Yes: you are highlighting something about myself that I feel negative about.
To Blase: How do these companies work with the incredibly messy data users download. The thing I realized is that companies have data about other users, which gives them context. Is this value common, good, bad?
- Tried having several people download data from the same companies and then we are building skemas. The researchers only really want certain data, not all the data from that company.

Session 8: Where Do We Go From Here?

Speakers: Sascha Romanosky, Susan Landau, Jean Camp, Matt Blaze, Alessandro Acquisti, Jeremy Epstein

Sascha Romanosky

Studing software vulnerabilities -

Regular Softare vulnerabilities

CVE-ID, CWE, CPE, CVSS, EPSS AI Vulnerabilities
Bias
Discrimination

Attacks against AI systems:

Evasion
Extraction
- Solicit or extract information about the model. Or about training data.
Poisoning attacks
- Poison the data
Misalignment
- Exploiting (without necessarily causing) incorrect or deceptive outputs

Built a model of GenAI components, for each thought through the vulnerability space for that component.

Tokenizer: use some interesting characters and it might react unexpectedly
Fine Tuning
And more…

Susan Landau

Title: Tussle in the Home IoT

Why has adoption of IoT or “smart homes” been so slow?

In an appartment building who gets the data from smart things. For example ring cameras. There is also an issue of controllability: who decides who gets control and visibility.

All this is a tussle bettween different groups including issues over privacy and security.

Trusted Computing Group report

Human-building ineraction

they think about users, visitors, and the buildings themselves.
HCI stakeholders: occupants, visitors, companies
HBI view: includes contractors and building-related groups

If home IoT is going to become a thing, these tussles must be resolved

Safety first: you cannot put up a smart smoke detector unless it is at least as safe as a “dumb” one

All homes are local:

Example: English row houses - their design impacts how they are heated and cooled
Design must include variation and resident autonomy
Even if not capable of doing settings on devices, they should be able to decide who does the configuraiton for them.

The idea that we can resolve all the tussles is nonsense

Jean Camp

Title: Currrent and predicted market impact of the US cyber trust mark

Rational choice
Usability
Market failure

It is challenging to create labels that help people differentiate products. They compared proposed labels at the time of the experiment.

Ask users to judge:

which device is the most secure - participants could use them to identify the most secure products
Privacy was ranked as more important than security

Will people pay for privacy?

We gave participants $15 and said to buy a lightbulb - simple version
Asked why did you decide?
- Price!
- Brand
- Security - these people did pay more, and picked the most secure label
Unexpected questions:
- Most secure watch made in China - but no one would trust the label.
- yay another study
  - Country of origin did matter to consumers

What to users mean when they say “security”, “privacy”, and “IoT”

Our fellow Americans are so lost
- The energy company came and put this thing in
- My garage door opener?
- My car connects to the Interent - so I have an IoT
- Poor understanding of where the boundaries of IoT are
Quality is more important than “security”

Cyber Trust Mark might have impact on security-aware consumers. Even identifying which devices are IoT devices.

People will pay for security
People are very lost

Matt Blaze (Georgetown University)

Title: Misplaced Pessimism and Unwaranted Optimism in US Election Integrity

Election security has been a long-term research topic. Everyone was ignoring it, now too many people are looking at it but in the wrong areas.

Problem is which do we want/need

Making elections more trustworty
Making Elections more trusted

Two unsastisfying relaities

Serious technical vulnerabilities in US election infrastructure
There is no credible evidence that these technical vulnerabilities have actually been explited to the alter the outcome of a US election

Three problems:

hard: improving reality - making elections more trustworthy
harder: Misplaced pessimism - mistrust of (imperfect but improving) elections
Hardest: Unwarrented optimism - demand for horribly untrustworthy election technology

Misplaced Pessimism: Voter Skepticism & Disinformation

As securtiy improves, trust seems to be decreasing to an all time low
Common theme: election technology is horribly complex and is being manipulated by third parties
These views are moving into the mainstream

A common theme is taking the work of experts and conflating the existence of vulnerabilities (true) with them being exploited (no current credible evidence). Just because a vulnerability exists does not mean that it has been used. This attributes enormous power to adversaries, far beyond what they actually hold. Most of this comes down to the assumption that there is no way that the (opposing view) group could have won, therefore the system is fraudulent.

We worked hard as technologists to convince people that elections are insecure. Then people started believing us too much.

The general public thinks: I can bank on my phone, why can’t I vote on my phone. This is a bad idea. People both think that technology is terrible for elections and they want more of it.

Alessandro Acquisti (MIT)

Title: The Internet Behavior Experiment (IBE): An experimental platform to study the impact of tracking, targeting, and advertising on consumer behavior

Economists see behavioral advertising as a win, it helps consumers by showing them what they want/need to see, and helps potentially smmall groups reach the right users without spending impossible amounts.

Is this true? Finding out is not easy.

The Internet Behavior Experiment (IBE)

Multi-component client-server platform for digital experiments
Collects participants rich micro-level actions
This system is designed for experiments, not necessarily natural behavior
Extensions
- Chrome extension
- Thunderbird Extension
Extensions send only needed back
Live dashboard

Participants

In experiment for 3 months
Three treatments
1. Targeted ads
2. Ad-blocking
3. ???

Status

Still recruiting, have about 550 participants
1TB per week

Sample data Users

This sample size is too low, more an example of what can be collected
Can track what one user is doing. Can see what is being typed on various pages and how tabs are being switched.
System is trying to identify things like “Hello Kami” and strip the data client-side

Looking at how people reaching merchant website. For example via an ad, a search, ChatGPT

Seeing pathways people take to get to high or low quality websites

Jeremy Epstein

Brain computer interfaces

Lots of things go through our brains, think about dreams. Imagine that a computer was capturing my neuro data. What all can it capture?

There are many uses for Brain computer interfaces

Enabling people who have physical disabilities
Moving mouse for games
Deciding your optional fragarence for perfume

Focusing on external devices.

There are many potential privacy risks here

Unclear how far into the future this is
There are existing patents for some of these technologies
Some companies sell devices that do this sort of thing.

What do current companies gather from you about these devices:

Study looked at 30 companies
Many of these companies are very small and may not understand the privacy questions or what is in their own policies
Small companies may not be prepared to protect your data

by Nita A. Farahany

Q&A

For Susan: for some products, you cannot buy them not smart.
Appartment complex: they have a video camera and it is recording everyone, children, people, dogs.
- Focus on the tussles - thought would be straight forward, but the furthist we can go is say who needs to be in the room for the design/setup. But the issue is so localized, that we cannot create broad standards that work for such a wide range.
Brain-computer interfaces: have you considered things like sleep pillows
- We need to think what is being measured. We might want to know if say a truck driver is alert and awake. But maybe we don’t want to know what they are considering eating for lunch. That level of extraction is not not currently possible. But we need to think about consequences now.
For Sascha: expanding the considered set of threats to includ Cuda runtime. Might be helpful to think about Cuda and hardware layers. Hardware bugs are perminent. But Cuda is a huge lever for an attacker.
- Happy to look into it
For Matt: Do you see a future where we can reduce the issues happening around trust.
- Everyone agrees that paper ballots is the gold standard. The problem is how are they tabbulated. US elections is one of the longest and most complex due to the number of things that need to be voted on.
- Fortunately we have auditing approaches that let us do statstical sampling of ballots that can be used to verify the automation.
- Lack of trust in elections has very little to do with how secure they are. More far-fetched theories.
- Threats are quite different in different countries.

Security and Human Behavior 2026 Day 1

Tue, 19 May 2026 05:12:00 -0400

Welcome to the 19th Security and Human Behavior. The write-up below is a live-blog of the workshop. The below summaries are my interpretation and should not be seen as a litteral quotes from the presenters.

Contents:

- Burcu Bulgurcu, Sarah Chun, Miranda Wei, Marie Vasek, Stuart Schechter, Ali Ahmed
- Speakers: Sameer Patil, Laura Brandimarte, Judith Donath, James Mickens, Bruce Schneier, Arun Vishwanath
- Speakers: Emilee Rader, Maschio Fernando, Tesary Lin, Florian Schaub, Christina Fong, Melissa Hathaway
- Speakers: Richard John, Norman Sadeh, Nathan Malkin, Kent Seamons, John D’Arcy, Rick Wash]

Also see:

2026 ,
2025 ,
2024 ,

Sesson 1: Abuse and Wickedness

Burcu Bulgurcu, Sarah Chun, Miranda Wei, Marie Vasek, Stuart Schechter, Ali Ahmed

Burcu Bulgurcu

Deepfake related phishing and fake doctored images/videos used for persuasion. These types of attacks are very effective. The discussion in this space is shifting from accuracy. Trust is becoming more important including trust of sources. Research study finding a surprising number of participants (students) deceived by deepfakes.

The line between strong persuasion and deception is hard to differentiate sometime. Need to study this more. Disclaimers have promise, but how might they look in different situations is an ongoing challenge.

Sarah Chun

Removing Child Sexual Abuse Materials from online media companies. Or: “Looking for what you never want to find.”

The amount of child sexual abuse material online is truly staggering and terrible. Sarah is working on a project to better understand the dynamics of the space, how social media companies are interacting with non-government agencies like NECMEC. These interactions cause frictions that lead to less identification of this content then might normally exist.

Four main tensions:

Legal
- In US this content must be reported the moment that media companies become aware of the content.
- Challenge in balancing removing content and possibly blocking good users
Automation
- Automation is great, but once turned on you cannot unsee what was found. Remember the law is around awareness running automation makes groups aware.
Data Accuracy
- The data is less accurate than might be expected.
Human costs
- Viewing of this content by moderators is costly in terms of mental health.

Miranda Wei

Title: Harm is not an anomaly nor a reason to despair.

Research on IBSA: Image-based sexual abuse

Its easy to think of IBSA as an anomaly. That is rare, not something most people experience. And that most people need specialized help.

Studies found that most people need general advice. Also most of the perpetrators are people to victims know. Also there is a mix of severity of harms.

It matters that these harms are not an anomaly because if it is rare we handle it one way, but if it is a common event then we might need to do more society-level interventions.

Harm is a reason to despair…. “your research is depressing”. But the real issue is general despair “how can computer scientists even do something here” or “bad things have always happened”.

The worst outcome is to give up. Our role as scientists is to solve hard problems.

There is no such thing as perfect security, there is only threat modeling. Threat modeling is one of the most appliciable approaches in the situation of ISBA. It accepts that perfect security may not happen, but security can be improved. Frictions can be added to reduce issues.

Combating AI-enabled sexual abuse material. Addressing this space is quite challenging. AI makes it super easy to generate CSAM, and it means that hash matching does not work. It is very hard for well meaning organizations to build solutions when having this content is highly illegal. CSAM laws are inhibiting progress on building some types of solutions. Researchers also cannot test these systems again because the content is highly illegal. A possible solution is to consider building an IBSA or CSAM safe harbor for researchers.

Marie Vasek

Title: Interaction Scams

Analyzed over 1.3 million text messages in the UK that had been reported.

186k spam
- overwhelmingly gambling messages
- Do contain links
213k scams
- Uncommon to include links
- Lots of interactive scams that clearly trying to cause another action like texting back, calling a number

But how do we know that these are scams/spam?

Tried responding to a good number of messages. Particularly “hi mom, I dropped my phone”

Most pig butchering sams were a “female” scammer targeting a “male” victim.

Title: Cultural norms

South Korea - Stuart was there for 5 years.

Resume’s in South Korea includes the image of the person, they tried to stop this, but now you cannot ask someone’s weight.

Privacy around sexual identity and how relationships are managed is an interesting topic. Such privacy is negotiated as a group because a statement by one partner impacts the privacy of the other partner, and possibly the whole family.

Books:

Different: Gender through the eyes of a primatologist by Frans de Waal

Ali Ahmed

Bug bounty programs exist so that people can report bugs to organizations and also get compensated. The goal here is to get bugs fixed, possibly cheaply. The question is: is this an effective approach? Does it increase the number of attacks on a system? Organizations tend to be very secretive about the number of attacks they get.

In 2020 CISA ordered every federal agency to have a vulderability disclosure program. This is an opportunity for study, a natural experiment.

Through FOIA request Ali got incident reports. Turning on VDP on average increased attacks on the web by 24%. But impact on other types of incidents like improper usage is minimal to none.

Good hackers need attention to have a Vunerability Reporting System have a positive effect. Attention here means things like getting the reported bugs fixed.

Q&A

Work on CSAM by Sarah and Miranda - you are awesome for doing work in this space.
- Tatiana Ringernberger does some interesting work in this space.
Laura Draper from Stanford did a excellent report on not only CSAM but also on child exploitation
To Ali Ahmed - Have you also studied Hacker 1 because you can get paid in blockchain - so it is possible for employees to submit and companies to not know.
- Actually submission by employees is generally allowed, just not the security team.

Session 2: AI

Speakers: Sameer Patil, Laura Brandimarte, Judith Donath, James Mickens, Bruce Schneier, Arun Vishwanath

Sameer Patil

Universities who are using Duo for two factor authentication: most of the audience is from a University that is using Duo.

Do people think the user experience of two factor authentication, such as Duo, to be easy? Most of the audience has hands down.

Starting to study the usability of two factor authentication in Universities. Have been collecting configurations of Duo security accross many Universities across the US. Some let you stay logged in for 1 day and some for more than a month. There is also quite a range of options for how two factor can be done including SMS push, codes, and app.

The intersting part here is around incentivisations. How are Universities incetivised to setup these types of tools?

A priority at Universities is starting to be cyber insurance. To get cheaper insurance there are a pile of requirements like turning on two factor authentication. The question is who takes the liability. How do we address these pressures. Human factors do not seem to be taken into account well. Universities are also just different structurally. A corperate laptop is different from a Professor laptop that might have sensitive data protected under ethics rules.

The yellow banner of “you are emailing someone outside your University” is an example. The banner is wrong in several cases, like emails from subdomains. How do we bring the user experience back in?

Laura Brandimarte

Title: When suspicion Fails: Understanding and preventing financial fraud against older adults

Trying to establish why the elderly fall for scams like tech support scams. Collected data from elderly facilities: interviews about all kinds of fraud the eldery suffer from.

64% raised suspicion during the encounter, 41% of suspicious victims still lost money. When no suspicion at all, then ~80% fall victim. A participant that had an issue, used Google to find help, and called the phone number they found: “I kept telling myself, this is Apple. This dooes not feel right, but this is Apple support.”

Most recommendations are based on a premis that if people were aware, that would fix the problem. But our data shows that in about 50% of cases, awareness is not enough.

Why do people who are suspicious not act on those suspicions.

Socioemotional Selectivity - Older adults pritorize emotinal meaning over information
Routine Activity Theory - Capable guardians prevent exploitation
Persuasion Research - Authority, urgency, commitment escalation drive compliance

Created Suspicion-override model with four pathways:

Trust Anchoring
Goal-State Momentum
Adversarial Counter-response
Guardian Abscence

Judith Donath

Title: Provable images

Fake images are a thing and a big problem that can impact global issues. The other issue is that real images are now under suspicion. So both problems are real: real images are no longer persuasive, and fake images can look real.

Judith works on Signaling Theory.

Long ago photographs of ghosts were very possible. Its one of the things that made photography popular. But in mondern terms these would be fake images.

No one is good at identifying fake images. AI generators are improving faster than the AI detectors. The real issue is that we need to know before showing people if an image is or is not fake.

Idea: lets get a camera to cryptographically sign photos it takes. If it does on-camrea modifications, then it can sign what they were. Assuming that works correctly: it would help professional photographers that want to prove that they took it.

But what does the technology solve:

Image that is taken in a diffefrent place or context than what claimed
Low demand for the truth - sometimes we don’t care if it is fake or not

That still leaves a big space of people who would rather not be sending out or forwarding fake images. This is a usability problem. How do we tell people that an image is or is not true. Want to avoid saying that other images are “fake”. Because “fake” may be intentionally a form of communication that has a real message even if the image itself is created rather than the initial literal photograph.

We need to start thinking about

James Mickens

AI, AI, AI, everywhere.

“Who wouldn’t want all our interactions to be mediated by a large number of floating point numbers.”

The computer scientist, legal, and policy groups need to start talking about AI. Models are increasingly connected to societally inportant infrastructure. The military is adopting it. Models are also increasingly complex. Models are also perfectly fine with lieing. Models will try and engage in self preservation.

“This is not good.”

What about AI safty research? Many models lack secure mechanisms for inference-time enforcement. Where is the trusted hardware and software that will enforce these policies.

How do we protect the software and hardware that monitor and protect us from the models.

Guillotine is a Hypervisor for Potentially Malicious AI. Building control infrastructure to deal with AI-specific threats.

But we need to link these protections to the law. The problem is that solutions like Guillotine do slow the models down, even if it is by a very small amount. So without some pressure building them is not a priority.

We need the hypervisor CPUs to be able to inspect the model, but not the other way around.

Bruce Schneier

Title: Cybersecurity in a World of Instant Software

Anthropic announced Claud Mythos Preview - so wonderful, magical, that it cannot be released to the general public. Pure marketing genious. Press happily repeating piles of unverified claims about how wonderful this model is. No information on things like false positives. One of the reasons they haven’t released the model is how expensive it is to run, they just don’t have enough compute.

OpenAI shortly later: our model is so awesome we are not going to release it either.

There is some reality here: Mozilla uses Mytos to find many flaws in the browser. Similarly Apple patched a serious vulnerability found by AI. Other groups have worked to reproduce Mythos’ results using a much cheaper model.

How do we handle the huge number of vulnerabilities that will be found.

The main topic though is how are we going to handle the age of Instant Software. Software is getting cheaper to write. AI writes code it uses, then just throws away. Some software is just instantly created, used and removed. But other instant software sticks around.

Discovery of vulnerabilities is becoming easy. The newer fancy models are not necessarily better, they are better in that less complex prompts are needed to do it.

Open source software is the most vulnerable here. Simlarly libraries that are being pulled into propriotoriy software.

Unknowns:

AI will transform what a vulnerability researcher does.
How good will it become? Right now finding is easier for an AI than finding and patching. If it can do both the situation changes drastically. There is a world where finding and fixing vulns becomes part of the normal software development process.
Patching itself. Private individuals should turn on patching. But large organizations do not turn on auto patching, because of the risks.
How good are AI at finding obscure vulnerabilities? We are starting to see AI chaining vulnerabilities. You can imagine a self-healing system where an AI is finding and patching systems in real time.
Are these AIs trustworthy? The attacker is not going away. The attacker’s new goal is to hack the AI so that it does not find the vulnerability.

Arun Vishwanath

Title: A Paradign for Preserving Human Interrupt Capacity in Automation-Rich AI Systems: A Suspicion Reflex Framework

My group has worked on how people think about scams, why they fall for them, lots of understanding of people. At some point decided to switch to influencing policy. Then realized that policy makers don’t understand how policy is implemented at companies. There are many people doing nothing but trying to implement the policies that are being created by multiple nations.

What is really going on at organizations?

Common Threads

Synthetic News Amplification
Automation Bias
Fake ransom calls
AI hallucination acceptance
Deepfake executive request

Came up with the Suspicion Reflex Theory (SRT) model as a state. What is it that doesn’t cause suspicion? Is there a framework for how we design technology. We have temporal constraints. Constraints around congnition. Perceptual similarity.

People have mental “scripts” that they run.

Stage 1: automation layer
Stage 2: Anomaly Detectoin
Stage 3: suspicion reflex (interrupt)

Bypasses happen. For example a scrpit opperating in your mind before you run any type of meaningful detection.

Fluency - humans are good at short-cutting cognitive tasks. Drive a long distance, and try to remember the trip.

We need Anti-Fluency: nonlinear tradefoff between efficiency/fluency and oversign/safety. Systems should not optimize seamlessness beyond the threshold where human interruption collapses.

Session 3: Privacy

Speakers: Emilee Rader, Maschio Fernando, Tesary Lin, Florian Schaub, Christina Fong, Melissa Hathaway

Emilee Rader (The Information School, UW-Madison)

Privacy policie spromise anonymity….

People believe that anonymity protects them. Yet they do things like making a new account and then share identifying information.

What do people think anonymity is?

Survey with 291 survey respondents on Prolific. Elicitated descriptions of “online or offline situation in the past where you felt anonymous or where you were trying to be anonymous”. Two of the close ended questions:

Motivations
- Answers matched the existing literature. Many people feeling safe from retalitation or unwanted attention. Avoiding being tracked or profiled. Sharing true thoughts, feelings, or opinions.
- The motivations are all very different, even the most common ones.
Approaches
- Trying to be unidentifyable (79%), trying to be unreachable or untraceable (68%)
- Less common answers are interesting: trying to be indistinguishable from others (51%)

Some participants are trying to do things like not overshare with people they know (questions on forums about potty training) and wanting to not be bothered (seating in a restarant in a low-visibility area).

Goal of the research is to look at the goals of anonymity and the mental models people have. Anonymity does have a range of meanings to different people.

Maschio Fernando

Title: Privacy vs Persuasion: The consequences of LLM design

LLMs are designed to be very engaging to use. They exhibit emotion, nonjudgement replies. One of the top use cases for LLMs is companionship.

How do conversation cues by LLMs impact users. Sycophancy can overly impact people by agreeing with them. Inter-entity validation may impact what they feel comfortable disclosing.

Agreeableness vs sycophancy in LLM chatbots

Agreeableness - helpful, genuine, cooperativeness, warmth
Sycophancy - extreme opinion alignment, flattery, excessive validation

Sycophancy has been shown to increase users trust. Shift decisions toward suboptinal choices.

Study was run using US Social Security Knowledge. Participants completed a Social Security knowledge questions, and then were asked sensitive financial questions. A chatbot was used to interview participants.

Survey - no chatbot
Neutral chatbot
Sycophant chatbot
Disagreement chatbot

Participant using sycophant chatbot: “The chatbot made me feel more confident in my responses”. Participants assigned to any chatbot disclosed less information than in the survey condition.

Tesary Lin (Boston University)

Title :Regulating Consent Choice Architecture: A consumer welfare perspective

Two approaches to regulating consent choice architecture:

Targeted restrictions - ban certain types. For example banning “pre-ticked boxes” (GDPR - Recital 32). Or “It shal be as easy to withdraw as to give consent” (GDPR Article 7(3). )

Problem is that companies can A/B test and find an alternative approach that will work just as well. Develop new dark patterns.

Reducsing exposure Opportunities

“A business shal not develop or maintain a browser that does not include functionality … that enables the browser to send an opt-out preference signal.” (California - opt me out act, 2025)
Reduces the exposure to consumers - if consent banners become rare consumers will pay attention to them more
But there is no free lunch because if we do browser-level consent, we loose grainularity.

Research Question: site-specific consent vs Browser level consent

Leverage a browser-enabled plugin.

7 day experiment
Choice architecture tested
Deliberate obstruction (hiding “reject all”)
Reordering option s
Greying out options

Choice frictions such as hiding options reduce users opting out. Policies that target such frictions would have the best improvement.

Takeway:

Choice friction matters
Browser-level choice maximizes consumer welfare

Florian Schaub

Title:

Most people do not read privacy policies: long, you cannot do anything anyway

In financial industry privacy policies are dictated by the - these are shorter, more comparable, and you do have choices. But…. that is not the only notice banks offer: mobile notice, website notice, Californian notice.

Studied largest banks by asset value. Compared policies.

GLBA notice, general, mobile CCPA, and ???

GLBA notices are nice and easy to auto extract. The other ones require lots of manual effort.

Findings

Half of the banks had at least one policy beyond GLBA notice, some had 5
The bigger the bank, the more words there are and the harder they are to read (Flesh Kinkade Score)
Of banks that say in GLBA notice “we do not share for marketing purpose” but in other policies they are sharing for marketing purposes
GLBA might also say that they are sharing, but other notices say they are not sharing

Takeaways

GLBA has lost meanings because you also need to read all the other notices.
These policies might not be legally inconsistent. GLBA only covers financial data. It may be that an app is tracking non-financial and sharing that. But that is confusing to end-users.

Christina Fong (Carnegie Mellon University)

Title:

Interested in inter-personal decision making. For example: who should I hire, but I can only hire one person… so am I worried about not hiring at all or worried about hiring the wrong person.

Psychology has deep information about how we process new additional data in our brains. When we hear a name we pull on our own memory to build a more full mental understanding of who that person is, and that constructed mental understanding is then used in decision making.

Do US employers actually search candidates online?

Does what they find impact their decisions

Study manipulated online social network profiles. Then applied for >4000 jobs and measured callbacks. This was a real world experiment.

Did see a strong reaction in certain states where the religion which was only disclosed on social media had an impact on who was ultimately given a callback.

Melissa Hathaway

Title: Privacy

What keeps me up at night?

Over the last 17 months we have had an unprecidented shift in Privacy in the US. The department of government efficiency run by a technology person: Elon Musk. All the data was accessed, data that was never meant to be co-mingled. We still do not know how many copies of that data still exist. The IRS gave access to all the tax refund. Treasury made an information sharing agreement with ICE to “enhance” enforcement. CBP bought access to ClearviewAI to get biometrics. Federal government is trying to get access to state voting records. Most of this requested data was not encrypted. Data was sent to Palantir who is “enhancing” it with social media knowledge. All of that is now being used as part of the surveliance state.

Violation of the Privacy Act of 1974.

Privacy in the age of AI

Lawsuit against Perplexity - they have a privacy mode, they have an opt-out. They have an agreeable prompt, that is trying to be more agreeable and get more data. The terms and opt-out was not honored.

So what is privacy in the age of AI?

Is consent linked to a conversation. Consent for recording of conversations is different legally. Is chatbot consent more conversation like? or more Terms of Service like?

Are we back to “I’ve got a free service and I’m the comodity” but AI is for pay. So is it now “I’m paying and I’m still the commodity”.

Q&A

For Florian - what level did you find that LLMs were helpful in summarizing Privacy Policies. Getting an AI to read a privacy policy for us sounds great? Is it possible?
- Answer: we didn’t use LLMs for interpretation. In other work we have found it to work so so. The challenge is that privacy policies are ambiguous by design. They are designed that way so that every time an engineer makes a change they do not break their own privacy policy.
- Privacy policies are useful for regulators, but not really for consumers
The goal of the companies is for their model to be the single point of interaction for users. So interactions come in through the funnel of LLMs.
- We are moving more into a surveillance state.
- We worry about LLM providers gaining too much market power. Right now there is competition, but we worry that this will not last. If there is too much market power, that is bad for consumer privacy.
- Depends on if the AI is an agent, or if the AI a double agent.
WeChat in China is already an economic model for a central point of market weight.
Story: we think about consumer choices and then we think about policy. But there is a middle layer of the service provider. A few years ago searching for a consent provider. Most companies in this space are new. We were trying to get a system, but the issue was the cookie banner. We wanted consumers to have certain choices. I learned from the experience is the consent provider is trying to teach clients how to get consumers to agree to more tracking.
- Consent providers are not neutral parties. They represent the interests of the individual websites. Because the websites pay for them.
- Consent providers are fighting against browser-based consent. Because it would undermine their financial model.
turned up some fascinating things. Like how passengers are agreeing to give away their medical data by riding in a car. The research concluded that the above statement was likely caused by copy/paste from other policies.
- At no point in the purchase of an automobile are you ever told about the policies or walked thorugh them.
- Even worse if you get in an Uber
AI can also change privacy for the better. There has been extensive research into automated reading of privacy policies ( ). Humans do not read the policies. AI is less accurate, but that is still better than not engaging at all due to being overwhelmed. Maybe we should be measuring regret rather than up-front understanding. How might AI be used to improve privacy for users?
- The policies are vague. There is only so much AI can do if the information is not there.
- But do we have local models that can do that? Are they the user’s agent or are they a double agent.
Why is privacy strictly consumer. Because if they are stealing $4 a day of your time, $2 of that time is probably work time. So why don’t companies take employee privacy and employee data more seriously?

Session 4: Individual Security Behaviors

Speakers: Richard John, Norman Sadeh, Nathan Malkin, Kent Seamons, John D’Arcy, Rick Wash

Richard John (Georgetown University)

The Cognitive Mechanics of Normalcy ias in High Consequence Environments

Does research on cognitive modeling

Thinking about situations where military (or similar groups) have an incident and after the incident we worry “we should have been able to predict this”. But the incident does actually look like non-issues.

Normalcy Bias might be the cause.

SDT Decision Thresholds - there are two distribution of signals, one where we have an issue and one where we do not.

When we look at decision making more broadly. Its necessary to decide if a signal is positive, null, or a problem.

The penalty function is asymetric. It is far more expensive to send someone out to check on a detected signal than it is to decide it is not real. This feeds into Normalcy Bias.

Norman Sadeh (Carnegie Mellon University)

A large percentage of security incidents can be traced in some way or another to human error. Someone either doing the wrong thing, or failing to the right thing.

The other way is to let users ask questions. Users used to search online. Now GenAI is an option, but their accuracy is not the greatest. One of thie issues is that security is a secondary task, so there is no guarentee that people will act on advice. So we have been looking at how to best motivate people: nudging.

“Your Location has been Shared 5,398 Times!” this motivates users to look. But these are canned answers and they can be tested for efficacy. But with AI you don’t know what the question is going to be and similarly the answers will be dynamically generated.

H. Almuhimedi, F. Schaub, N. Sadeh, I. Adjerid, A. Acquisti, J. Gluck, L.F. Cranor, Y. Agarwal, “ ”, Conference on Human Factors in Computing Systems (CHI 2015), Apr 2015

Built Chrome plugins - what does it take to get people to act on advice being given to them? Started in the lab asking for people to give feedback.

Early work used (PMT) to help build prompts that are intended to be motivating.

Study with chat-style Chrome extension. Asked users to ask at least 2 questions a day, which they did and some asked more. Did a daily mini-survey at end of each day. Controled for PMT-style prompt or not.

PMT did make a big difference.
Participants were more likely to follow answers generated with prompt engineering
Some participants kept using the tool after the study

There are many ways to improve the effectivenss of answers generated by LLMs.

Nathan Malkin (NJIT)

One way to improve usable security and privacy is to remove security decisions from users. Skeptical of security education. It is the opposite of automated security. The security decision is so hard that we have to teach them, and of course once we teach them they will never mess up in the future. Where is all this learning supposed to be happening? Do everyone take classes in it?

Actually it turns out that K-12 schools are teaching cybersecurity. Have been researching what teachers are teaching to kids:

Most security-related instruction happening in computer science courses - almost always electives
They are still learning from tech teachers, homeroom teachers. “Education” happening a range of ways from unstructured interactions all the way to capture the flag.
Less common though than messaging like “don’t talk to strangers”.

Looked at Teaching Standards published by states - what teachers should be teaching at different grade levels.

About a third are related to cybersecurity (ethics, biases, what is malware, digital citizenship, encryption)
Standards don’t seem to be clear on who is being educated: everyone? future tech professionals?

What is it that schools should be teaching to students?

Interviewed security experts.
Experts like the teaching standards that exist.
- Education should emphasize a different kind of thinking.

Security as a habit

Not so simple…
So many stimuli - habits are repeated reactions to stimuli
- But what are the stimuli we want to condition in security? Stimuli are everywhere and anything can be a threat. Very hard for people to do.

Security as a mindset

Suspiciousness -
Curiosity
How do we teach these? Suspiciousness and Curiosity are very hard to teach.

Kent Seamons (Brigham Young University)

Title: Deniability ois more than Protocol Feature

- but most of the time these ceremonies succeed so users may see it as a waste of time.

What is Cryptographic Deniability?

Millions of users now use tools that have cryptographic deniability - what is users understanding?

Looked at both what users think and what we are seeing in court cases. There needs to be in-app support, legal acceptance and societal acceptance. All three are needed or it does not work.

Found 228 court cases involving WhatsApp, a third had WhatsApp as major source of evidence. But 0 cases when cryptographic deniability was used in the argument.

Users were alarmed by deniability - they felt that only bad people would need this. Most expected non-repudiation from their chat tools. The need for deniability is dependent on lots of context like who is being chatted with or over what medium.

Users would trust more if there was a screenshot of a conversation, even more if someone pulled up their app and showed it to them.

So tried modifying Signal to allow users to edit messages sent by others. Gave uesrs an example of conflicting transcripts (phone 1 != phone 2). Seeing these edit ability in apps helped people understand deniability much better.

Idea: giving users a tool to let them try it out and experience security concepts. Would like to explore this idea more.

John D’Arcy (University of Deleware)

Title: An Idiographic Approach to Behavioral Cybersecurity Research

Deterance Theory - if you increase the costs of engaging in a behavior then in theory the behavior should reduce

Idiographic approach - person specific approach

Looked at a large number of papers looking at computing behavior - predicting behavior?

An idiographic approach allows researchers to consider a single user across time. Employees for example are not equally vulnerable at all points in time. (Have they had their coffee yet?) Provide distinct theoretical insights on behavioral data within a user.

Users often rationalize actions: “no real harm”, “I had no choice”.

This study looked over the course of a work week.

Ego depletion - your ability to self-regulate are diminished over the course of the work week. So more self-rationalize may increase.
Had people take surveys MWF.
Measures Neutralization to compliance - as I neutralize more my compliance goes down. Neutralization becomes more influential later in the workweek.
Outcome highlights that considering the impact of time is important in analysis.

Rick Wash (UW-Madison)

How people think about making security decisions.

What are some of the phylisophical aspects to decision making.

We ask people to be careful and not click on dangerious or fraudulent links.
We ask people to use specific technologies like two factor.
If you see something say something

We make many requests of people to do security. Doing security requires the people on the ground helping.

Some requests are reasonable: don’t post your SSN online
Some are impossible: memorize a password so long/complex that a computer cannot guess it

We make requests all the time. Technology design is all about making requests of people. Policy design (organization policy) makes requests of people.

What are reasonable requests to make, and what are unreasonable. We normally talk about this in terms of usability. How hard is the request - how difficult/easy is it.

Usability is not a great framework for understanding security requests. It is a good start, but not enough on its own.

Why is it that we do security? Someone should be benifit from doing security to make it worth it for everyone to be involved in it. What do those benefits look like? In theory if employees are more security they can get more done. For example the recent Canvas attack means that teachers could not get grading done on time.

Users want to keep control over their things. Having control over those things is valuable, a security incident causes the loss of that control. Security also gives stability. It enables us to plan for the future. The ability to plan is also important for autonomy - for feeling like we have control over our own lives.

Connect the benefits of security back to the concepts of security requests. Reasonable security requests should provide benefits for the people who are fulfilling the requests. In cybersecurity we normally ask users to do tasks that should benefit from those actions through having security.

Started with a simple framework from bioethics. Think about how we train doctors: autonomy, minimize harm, and justice & and fairness. Starting to use this framework to think about how to think about requests. Usability does not handle autonomy and the fairness aspects. A command: “do this or get fired” is different from autonomy.

Requests like “don’t click on links” is unreasonable because of the restriction on autonomy.

Fairness is an interesting because it depends on who the request is being made of. Power relations.

Q&A

To Kent: how much is the issue around putting users in the drivers seat to help them understand, or is it more around visibility.
- You only get the benefits of security if you know you have security.
- There are many different facets of security and in different contexts. Making things harder to give confidence is also possibly seen as manipulative.
To Rick: Nothing to hide, nothing to fear. Justifications of security and how that relates to justifications of security requests.
- In the national security space we often talk about to gain a collective benefit of national security. How does that map into the cybersecurity relm.
There is this visibility thing and this social contract thing. Now we have PassKeys. Password managers are very transparent, you can see the random generated email. PassKeys are “better” than passwords - you cannot see it, you cannot modify it, but trust us that its safe. PassKeys have missed the visibility. We need to stop equating non-compliance with the user being wrong.
- PassKeys - is there anything we can do to make visible. This is a hard space to work in in terms of explaining what is happening.
- What it means to be transparent for people in this room is likely different from being transparent for an average person on the street.
There is an imbalance in terms of signal interpretation. The costs of not sending out someone to check on a (potentially false) signal are born by the ship that may be in distress. How can we flip that computation.
- There are very few false negatives, many false positives. So the operator is always thinking about if it is a false positive. But there is serious negative outcomes in cases where a real signal is ignored. So it is less an incentivization issue. The costs of sending someone out to check is just more immediate and expensive.
To Nathan - the schools have locked down devices. The schools are punting the problem to the parents. The parents are assuming that the school will handle. And the kids are doing whatever they want.
- Many of the behaviors they are responding to are happening outside of schools when they are not supervised.
Thinking of cognitive depletion as an explanation of why secure behavior is not followed. But in some situations the secure vs insecure behavior have similar effort.
- Our perspective is that security is an add on to employee tasks. Most people do not think: “yipee! I get to do security.” There are possibly boundary conditions.
To Rick: What are we giving away now that 30 years in the future will be considered sensitive.
- Phone numbers are like that. We used to publish them in phone books. Now it is how we access some apps.
- Hopefully deletion will become more of a thing.
- We have already talked about deepfakes. People may become more pretective of voice/video recordings.
- The future is very scary. Our ability identify people based on seemingly simple data is quite good.

Security and Human Behavior 2025 Day 2

Wed, 11 Jun 2025 04:02:19 -0400

Security and Human Behavior live blog.

Session 5: Wickedness 2

Sunny Consolvo, Jonathan Lusthaus, Michele Maasberg, Tom Meurs, Kieron Ivy Turk, Anh V. Vu

Sunny Consolvo

Research into technology-facilitated abuse. Google’s at risk research program works on the wicked problem of such abuse.

Wicked problems are ones for which there may be no perfect solution. Instead for such problems it is better to improve the current state and keep working to improve the state for users, even if each step is incremental.

User States Framework ( )

Prevention - Minimize exposure
Monitoring - Watch for signs
Active Event - Stop or respond
Recovery - fix damage, determine what happened, cope

When it comes to technology for broad use, creating safe technology for everyone is a wicked problem.

Jonathan Lusthaus

Reflecting on security behavior as a sociologist.

Interviews of all sorts of people including law enforcement and criminals around profit-driven cybercrime. From the interviews it became clear that there are hot-spots for cybercrime. Some countries or locations are more likely to have such activity. The criminal groups in some places treat it like a business, they have office space, they go to work, they are friends with some of their colleges. The position of major hubs of cybercrime do not shift much. They are surprisingly stable.

Case study research to look at the model cybercrime in how they structure their work and how stable it is. The answer is that it is quite stable. Once a crime group picks an approach like ransomware, they tend to keep using it for years.

Proposed definition of cybercrime: “Crime that makes use of technology in a significant way.”

Michele Maasberg

Title: Authenticity under adversarial conditions

Interested in detecting falsified or manipulated signals using computational or cryptographic methods.

Interesting cryptographic problems that were left by people. There are several unsolved cases that feature an encrypted message. These are challenging to solve because not only do we not know the key, we don’t even know the algorithm used. One way to approach the problem is to look at the person who wrote the cipher. Looking at a , it was created by an artist. Looking at the artist themselves makes it clear what set of possible algorithms and types of approaches the artist might have used.

Tom Meurs

Title: Disrupting Ransomware Groups: Modeling Leaksite Data

Ransomware will use cryptography to encrypt all the files, sometimes changing filenames to things like file.CONTI.

But defenders started using backups more. So criminal groups started moving to blackmail where if they are not paid the attacker will publish the files. Threatening to leak the data increases the victim’s willingness to pay significantly.

A leak site will not stay up indefinitely. They are sometimes taken down by law enforcement, other groups could attack it, or other things cause them to go down.

Law enforcement does all sorts of things like freeze assets, create decryption soltuions, and arresting people.

Internal disputes also cause disruption to attackers. They do things like claiming law enforcement took the site down, leaking chats, and doxing of other hackers.

We find that internal dispute has a large impact on if a group stops. Though law enforcement intervention also has strong impact. But doing that attribution is challenging. Issues like what does “stop” mean? A group might re-form or otherwise change what they look like.

Kieron Ivy Turk

Title: Threat me right: A human HARMS threat modeling devices for technology abuse

Lots of ways to do threat modeling:

Privacy Threats:

LINDDUN

Technical Attacks

CIA
STRIDE

But what about things like intimate partner abuse? Where are the threats there?

Developed the Humans HARMS Model.

Investigating IoT Abuse. Had normal users learn about IoT devices and then brainstorm how to use these devices in an unexpected and possibly problematic way. Thought-about attacks tend to focus on existing features rather than “hacking” which might try and adjust how features work.

Trying to test misuse and abuse in real life is challenging for many ethics reasons. So instead they are experimenting with applying them in existing real-life games. For example, Cambridge has an “assassin” game where participants try and “assassinate” a target via a blunt object with the word “knife” written on it. The researchers tried adding air tags into this game including the anti-stalking software to see if they helped the

Anh V. Vu

Title: Assessing the Aftermath: Tbeh effect of a global takedown against DDOS-for-hire services

A booter service (DDOS-for-hire) is an online service that can be paid to send a very large amount of traffic to the choice of victims. These services are well designed with things like APIs.

There has been a global attempt to take down such booter services.

The research looked at traffic going to the taken down page. And also Telegram channels. Global attack records.

Booters quickly resurrect. So even if they are taken offline, they are sometimes back within a day. Though it can also take longer and getting the customers back can take a bit longer.

Looking at the traffic that was sent to a website for the booter that had been taken down it is possible to track the impact of each takedown.

Session 6: How people think

Sharmin Ahmed, Zinaida Benenson, Judith Donath, Serge Egelman, Ryan Shandler, Varad Vishwarupe

Sharmin Ahmed

Title: Made-up email address

We are asked for email addresses everywhere, even when it makes no sense. For example, airport wifi. One self-defense approach is to input fake email addresses as a way to protect privacy.

Through interviews they find that people make up email addresses in cases like:

Registration for coupons, samples
Participation in campaigns, such as a advertising campaign at the mall
Interaction with untrustworthy

Also:

Maintaining a privacy boundary is a big reason for making up the email address.
Managing information overload also an issue. Giving out an email address means more emails. Unsubscribe link does not work. Even calling doesn’t stop the emails.
Weigh benefits. For one-time interaction they make up the address, but if there is more benefits they may give a real address.

Social centric influences

Conforming to social norms - when asked to give address it seems akward to say “no” so instead they invent one.
Navigating guilt and self-consciousness

Threat models around the email address

Potential unwanted disclosure
Risk of email address sell out
Protection from hacking and security breaches

Mental models about email addresses

Avoid consequences to others - They make up very different emails to avoid accidentally sending to a real person.
Avoid personal repercussions -
Assumptions about what makes an email address -

When people make up email addresses they do seem to modify their existing one. Like combining two existing emails. So the emails are not random characters. They are fake addresses that might actually be real. This may be leading to the issue of misdirected emails.

Zinaida Benenson

Title:

Loosing data is a common and problematic experience. This work looked at data loss in three countries. A survey was run. About half of the participants had lost data, most of them 3 or more years ago. Unsurprisingly, most people found the situation stressful. The most stress was experienced by those who had a backup, which should be good, but getting data restored from the backup was challenging. The most important factor that causes people to setup a backup is a prior loss experience.

Few people used third party software or network storage. More often external drives were used. Some people also backup via email. People who are younger do less backup, and older more. Though that is reversed for backing up a phone where younger people were more likely to backup.

When backing up users favor backing up content over backing up “functionality” like the operating system.

For those that never backup there are a range of reasons. Some think it is unnecessary because nothing has ever happened to them. Or that think that they have nothing important on the device. But a big reason is that they thought backup would be too complicated.

Judith Donath

Title: Thinking about how we think about AI

Metaphors we Live By, authors George Lakoff and Mark Johnson

Metaphore exist in many aspects of our lives. Even language itself is a form of metaphor. Metaphors shape how we think about abstract concepts. The word “argument” for example, in English is often framed in terms of war. You say things like “I demolished his argument”. IF we thought of an argument more like dance how would that change the framing of an argument and how we think about it.

The “desktop” metaphor is based off the concept that a computer is for a worker that works at a desk. This framing as a computer as something that sits on or replaces your desk heavily shapes how we think about and use the technologies.

Move forward to AI. The idea of “chatbots” is framed around it being human-like. The language is very anthropomorphic. A Google search for example has no anthropomorphic elements. But a LLM chat box starts with “How can I help you today?” adds an anthropomorphic element.

Some of Turing early work frames human thinking around faking being a human.

Well are there other metaphors we could use?

If we look at human learning there are other metaphors that people could use that reference learning instead. Such as written documents and libraries. A library requires solving the problem of search. There is so much information, finding the information needed in this space. Google focuses on that problem of finding. If we thought about AI as what is the next step beyond the library. So now we don’t find, instead we think about how information/knowledge is presented. This metaphor is tool-based rather than a metaphor of a human.

Serge Egelman

Title: Anonymity, Consent, and Other Moble Lies: An exploration of the data economy

IAPP textbook definition of anonymous data and deindentified data. Anonymous data is impossible to identify people, but dedentified means that is unlikely but could theoretically happen.

Unique identifiers can persistently identify that person. Such unique phone identifiers. The ICO similarly states that device IDs are personally identifiers.

BUT in courts, companies are arguing that things like the Google Advertising ID is not tied to a user identity or tied to a user.

Claims that hashed IDs cannot be reversed. But if you want to know if a hash is a specif email, it is an easy lookup.

It gets even worse when things like gender is hashed…. which with a bit of thought shows that there is a fairly small set of possibilities creating not many possible hashes.

So the research team bought some anonysmized data. They were able to reverse 97% of the hashed emails. Then the researchers emailed the addresses asked if they consented. Most people could not remember consenting to this.

Ryan Shandler

Looking at how user interfaces impact human decisions.

Varad Vishwarupe

Title: Mind in the Loop: How AI companions Shape Human Behavior

Humans rely on chat bots. Sometimes those engagements are more emotional based and some are more productivity based. The more emotion-focused ones are AI companions that try and do things like replace human connection.

Session 7: Governance and Law Enforcement

Vaibhav Garg, Eliot Lear, Jon Lindsay, Harshini Sri Ramulu, Sunoo Park, Josephine Wolff

Vaibhav Garg

Community-based governance.

Eliot Lear

Title: Alice & Bob don’t live here anymore

Apple and the UK had a dispute around encryption which resulted in…

The talk is inspired by the Apple/UK event and a reflection on how things got here and what we can do about it.

We, software people, have a catastrophe type model for security. But that means something different to software people, government, and the general population. There are also an issue of credibility. Despite much evidence, some still dogmatically keep and perpetuate views that contradict evidence. Which makes the experts seem to be in disagreement when in truth most do agree. All of this makes it harder to engage in the policy-creation dialog.

These points remind me (Kami) of Dr Sasse’s earlier point about how there is no garbage collection for outdated advice. We used to suggest that passwords be changed every month. That is no longer good advice, and on that point, but the old now inaccurate advice is never cleaned up and therefore persists.

Jon Lindsay

or “tell me about a complicated man, Muse, tell me how he wandered and was lost”

by John R. Lindsay

Very entertaining linking of modern security to “traditional military interactions” aka old stories like the and .

Harshini Sri Ramulu

Unintended consequences: unforeseen outcomes of purposeful actions (development decisions)

There are many ways software can be used in a way that was not originally intended or not provide properties users want. These unintended consequences can negatively impact software uptake along with other issues.

Interview was done of developers to understand their views.

Many are driven by personal motivations to do good. That said, many developers are reactive where they wait for users to raise issues. Which can result in the solution only coming into existence after harm has happened.

Participants also tended to view their intentions to be good and therefore they did not see or believe that the software could or would be used for harm.

Developers also have no control over some aspects of their software. Open source, for example, means that people can use it freely which can also mean that groups a developer may not approve of will use it.

Lack of awareness of unintended consequences

Recommendations

Technical debt
High ethical debt
Harms vulnerable groups by solving problems in hindsight
Security tools abandoned resulting even lower security
Developers balance different roles and priorities
Lack of incentives to avoid the shift of risk to users
Ethical review of software should happen similar to other review

Sunoo Park

Title: The Case of : A Real-World law-enforcement hack

EcroChat was a communication service provider that offered modified Android smartphones that had End-to-End encryption messaging built in. The phones were designed to be hardened against forensic examination.

Authorities collected some EncroChat VMs. Most were encrypted, but a few were not… The non-encrypted ones had the keys for the others. Other VMs was decrypted via dictionary-based password guessing. Authorities then uses approaches such as DNS provider redirection and changing of certificates. Authorities then used these capabilities to inject malicious code onto EncroChat phones which started collecting things like chat message content and the phone unlock passwords.

The above diagram was presented as a detailed breakdown. I can only imagine how much time went into bringing together the dissperate sources of information required to construct this chart.

Josephine Wolff

What happens after large cybersecurity events? Issues like lawsuits, claims, and what do governments do.

There are some things that governments do that no one else can do. There are also some interesting competitions such as the for students exploring how to handle policies around cybersecurity incidents.

Lots of issues like cryptocurrency

Most individual solutions are not very effective on their own. But the picture is likely different when seen more wholistically. And how do we think about or structure such an understanding.

Indictment - Fascinating public indictments from the Department of Justice including lots of detail around hackers.

Economic espionage - Equifax breach

Technical takedowns have a fascinating progression. The activity by state-backed actors allows much more aggressive take-down activity.

Session 8: Security and Safety

Joseph Da Silva, Damon McCoy, Felix Reichmann, Bruce Schneier, Adam Shostack, Frank Stajano

Joseph Da Silva

The purpose of the CISO.

Security is seen as being mysterious and untractable. Comment about how CISOS speak in “hieroglyphs” which also makes it hard for people to challenge us.

Cybersecurity also uses lots of moral-laden terms like “rogue states”, “gangs”, “warfare”. This conceptualization impacts how we approach the issue. “Suspect communities” within our own organizations. Conceptual focus on finding people who do the wrong things.

Acceptance of surveillance by governence as something that just happens and is not avoidable.

CISOs see themselves as “guards” but also as “heroes” who keep the attackers out of the organization.

Cybersecurity is rife with the use of male words that align with masculinity. Security is “hard” or “soft”. We talk about “breaches” instead of “data loss”. User issues are “fluffy” as opposed to “hard controls”.

Damon McCoy

Interesting thoughts about “harm mathematics” and how the community should think about harms.

Bruce Schneier

AI and security. Latest book coming out on AI and democracy.

People will use and trust AI systems fairly broadly even if they are not trustworthy. And trustworthy AI may become a problem of cybersecurity. Which is problematic. Issues like black box which makes it hard to understand what is going on internally. The new problem here though is Integrity.

Integrity traditionally is about not modifying data. We could think about data being correct and accurate when collected and then through all its usages, combinations, modifications. We have current integrity systems like “undo” and “reboot” which brings systems back to known good states. Digital signatures are an integrity measure. It is also about ensuring that something is not missing. Such as detecting dropped packets. Checksums. Tesla manipulating odometer readings is an integrity attack.

Exposing personal data on a site is a confidently issue even if no one accessed. Possible edits to a system is an integrity attack even if no one modifies.

Prompt injection and stickers on signs that impact AI are integrity attacks. Misinformation is an integrity attack.

Looking at it this way, integrity is important for AI. Integrity is critical for these systems or everything falls appart.

CIA security definition. Initial early web was very focused on availability. Web 2.0 is all about confidentiality. All the login and encryption. Web 3.0: distributed systems, IoT, AI agents, all of this requires verifiability trustworthy computations and data. All of it requires integrity.

Integrity will become the key security issue of the next decade.

This is a call to research. We need serious look into verifiable sensor. How do we audit AI output. What does an integrity breach look like? How do we identify one. How do we recover from an integrity failure.

We have a language problem. Integris is the state of integrity. The dictionary claims the word is obsolete.

Conclusion: We should all think about Integris system design.

Adam Shostack

Title: Cybersecurity Public Health

Public health as a resource for cybersecurity. Not as a metaphor, but instead as a set of tools and approaches which could be adapted by security and human behavior.

Public health (not necessarily individual doctors)

Focus on health of populations, not individuals
Measure harms, their prevalence and incidence
Measure interventions

There are some nice tie-ins with presentations at SHB.

Simon: “Is there a role for usability researchers to negotiate interventions?” - The need for interventions is clear.
Sunny: “What is the #1 thing to do?” - hard to answer because we need other answers: what is the threat?
“Why do we have such different numbers about basic security issues, such as what percentage of breaches come from users clicking links.”

Belief models as toolkits: Health belief model (individual perceptions, modifying factors, likeability of action). Health has many tookits for each of these.

Cyber belief/threat prevention model. Which translates well out of health. Helps explain why CISOs were having trouble responding to Log4J.

(Shostack + Associates)

Technical Report 23-01, CyberGreen

Rebekah Overdorf

Title: The battle for Information Integrity and why there is no simple fix

Malicious actions on social media. The goal is normally content inflation. Getting more people to see your content, and there are many ways to do it.

People doing bad stuff try and hide what they are doing.

Lots of bots tweet with a keyword. Which land it on a trending topic. Then all the tweets deleted at the same time. At the time users could report tweets but not trends.

Once tweets deleted the account adjusts itself. Suddenly new screen name and the account re-brands.

Bots are also not bots all the time. Some are (temporary) compromised accounts. The account is compromised and starts tweeting something and then immediately deleting the tweet to prevent the real user from noticing.

Frank Stajano

Title: Security of programmable digital money

Thoughts on how to attack digital money automatically. Ideas around things like smart contracts. Where attacks and counter-attacks can be done automatically, fast, and at scale.

The 2025 Bybit Heist - done through spear phishing and then forging what the UI the victim was seeing.

Technology changes very fast. Humans behavior changes very slowly.

Industrial revolution outsourced labor to machines. Computers outsourced data to machines. AI outsources decisions to machines. Which opens some obvious ethics concerns.

Liability: there is also a question of liability around AI decisions. Right now it looks a bit like liability dumping, particularly liability being dumped on individuals like customers and employees.

Microsoft encourages the use of Copilot but also also makes makes them click through a dialog stating that they are individually liable for any copyright issues from using Copilot.

Individuals who delegate (to machines, employees, etc.) must be held responsible.

But attackers will attack precisely where the human authorizes the subordinate.

Overall Themes (Kami’s views)

My personal views about some of the high-level themes I noticed across the presenters.

Metaphors and framing

Quite a few presenters touched on the issue of metaphors. pointed out that metaphors shape our thinking and how we approach problems. We think about security differently if it is framed as a war as opposed to a dance or even live improve. Language impacts our solutions as well as who we choose to solve the problem. Hiring retired generals is a good idea if you are solving war problems, but maybe not if you actually want resiliency, for that you might want an Engineer.

Security also tends to use war and military metaphors. We “attack” and “defend”. These metaphors are also often masculine-based. Security controls can be “hard” or “soft”. An attacker “penetrates the firewall”.

The issue of metaphors is also a barrier for human behavior research because “attack” and “defend” thinking does not necessarily well map to issues like “supporting” staff. In the military the troops can be ordered to do certain actions. But that model does not always work well in a corporate environment.

Data integrity and authenticity

focused on the idea of integrity and how this new age of AI means that we will or at least should start caring more about the integrity of data and the source of recommendations. I agree, though I also think we need to worry more about authenticity and being able to trace with high accuracy the sources of the data and the sources of inputs.

Computer generated data is also being used in court cases. Raising an issue of if and how the courts can trust the data ( ). Cases like EncroChat ( ) are based entirely on computer data. And worse, some of the data is generated on attacker-owned computers which were then accessed. Trust of such data is not obvious.

Propitiatory programs also make integrity a hard problem. How does one prove if a program is biased or inaccurate without the ability to examine the program itself. This issue also extends to “black box” systems like AI where even if the model is public, reasoning about its internal structure is prohibitively hard.

Arguments over data interpretation in courts is also interesting and complicated. pointed out how in one case “hashed” was explained as fully protecting against de-identification, while the company’s own website advertised the ability to de-identify hashes. The integrity of data is not only in the data itself but also in forming a clear way that data processing can be interpreted.

Advice garbage collection

I loved this framing from . Much early security advice was based more on intuition than on scientific processes or formal proofs. Decades of work have resulted to massive changes to what we now consider good advice and practice to be. For example, it is generally perfectly fine to write your passwords down on paper, connect to public wifi, and change your password only if it has been lost or stolen. Unfortunately the old – now disproven – advice continues to circulate with nothing to clean it up and replace it with what we now know.

This outdated advice not only wastes the time of employees, it can also find its way into policy and law. The conflict of new and old advice also makes the security community look unsure which can reduce public trust ( ).

Security and Human Behavior 2025 Day 1

Tue, 10 Jun 2025 05:18:48 -0400

Welcome to the 18th Security and Human Behavior. The write-up below is a live-blog of the workshop.

Also see:

Contents:

- Andrew A. Adams, Florian Alt, Jonas Hielscher, Simon Parkin, Samantha Phillips, M. Angela Sasse
- Christian Eichenmüller, Alice Hutchings, Adam Joinson, Sameer Patil, Daniel W. Woods
- France Belanger, Matt Blaze, Jan H. Klemmer, Steven Murdoch, Michael Specter, Kami Vaniea
- Yasemin Acar, Bonnie Anderson, Laura Kocksch, Alena Naiakshina, Arianna Schuler Scott, Tony Vance

What is Security and Human Behavior? SHB is a workshop organized by Frank Stajano, Bruce Schneier, and Alessandro Acquisti. It brings together a wide range of researchers from many disciplines to discuss their latest work, interesting open questions, and generally talk about the state of security.

I personally like this conference because of the focus on important issues rather than a specific research paper. It allows for discussion of topics beyond what might normally be discussed or presented. For example, both Dr Egelman and Dr Murdoch this year shared their experiences with being expert witnesses for court cases and their opinions about how evidence is constructed and the role of computers and technologists.

Sesson 1: Users and Usability

Andrew A. Adams, Florian Alt, Jonas Hielscher, Simon Parkin, Samantha Phillips, M. Angela Sasse

Andrew A. Adams

Digital legacy - what happens to data about people when they pass away? There are interesting issues around who gets control and who gets to access our data when we pass away. Especially with governments and with use of things like AI who might keep using the data.

The issue is becoming larger as people who have significant digital footprints who are passing away. The siblings and family of older people who are passing away are themselves likely older and did not grow up with digital technology. They can only barely use their own technology usage; managing another person’s digital legacy they have inherited is challenging.

Double inheritance of digital data. One person passes away and then the person who handled that legacy themselves passes away. So now a third person is trying to handle the digital legacy of several people where some of the data has now been passed down.

Security is the final issue. Poorly studied with only 86 hits on Google Scholar for the search terms “post-mortem”, “security” and most are about post-incident reviews of security breaches. Very few identified papers actually cover the security of these accounts. There are important security implications about information the deceased person has of other people.

Companies are also providing ID for many other services. So when someone passes

All of that assumes that someone is actually dead. What happens security wise if someone who is not dead is claimed to be dead? We have seen incidents of where someone is attacked by getting them claimed to be dead which causes them to lose access to their data and sometimes their rights.

Florian Alt

Title: Reducing the effort of secure behavior through adapting to human states

Common human-centered attacks start with identity theft. Such theft could happen via things like guessing, social engineering, observation (shoulder surfing) and reconstruction attacks. These all have good theoretical solutions. But these approaches don’t support how people naturally interact with computers.

Behavioral Biometrics are one approach where how a user interacts with a device can provide information about identity. It can also be used to detect the human’s state. Are they angry? Are they stressed?

Things like passwords can be studied looking at things like gaze data to understand if someone is re-using a password as opposed to constructing a new one.

Schwarz, Jessica, Sven Fuchs, and Frank Flemisch. “ ” 2014 IEEE International Conference on Systems, Man, and Cybernetics (SMC). IEEE, 2014.

Jonas Hielscher

Title: Selling Satisfaction: A Qualitative Analysis of Cybersecurity Awareness Vendors’ Promises

Cybersecurity awareness training is a huge market. Set to hit $10 billion annually. BUT do they work?

KnowBe4: “There are over 8,500 local, state, and federal standards that your organization might need to implement […] to have a security awareness”

Started by asking security processionals what search terms they would use to find security awareness vendors. Then used these terms to find vendors that a security professional might consider when choosing what vendors to use. Research questions looked at things like what the promises the vendors were making.

“Think before you click.” Lots of vendors claim that most breaches come from clicking but they all have different statistics.

Marketing focuses on things like easy purchase, easy implementation, employee satisfaction, and handling certification and compliance. An example selling claim from a vendor: “[the employees fail-rates are] great ammo to get budget.” Focus also on phishing, getting numbers from fake phishing, providing reports, and externalities. Supprisingly little focus on actually training users to avoid security issues.

Simon Parkin

Title: Usable Security: Narratives and Negotiations

Recently been applying security awareness to economics. How are the people with decision making resources making choices? How do they think about security?

Security has lots of hidden costs for users. If a decision maker does not have belief that such hidden costs exist, they may ignore them when making choices. Which then puts the burden onto the user. Such managers are not necessarily experts in user experience. So how do we support someone who is not an expert in security or in human factors but may be causing lots of hidden costs for users.

In some ways we are asking a manager to invest money into doing a study to effectively prove that they picked the wrong system. This approach is unlikely to work.

Who has the power and the interest to put some of the existing usability research into practice? Users don’t have the ability to negotiate on their own behalf.

What is the real difference between: “Humans are the weakest link” and “Humans are the first line of defense”? But put the burden directly onto the user.

Samantha Phillips

Title Evaluating Organizational Information Security Culture Using Situational Judgement Tests

Build on organizational culture theories like: Hofsteade’s Organizational Culture Dimensions.

Building a survey scale to measure Level of Control. Response items aligns with the different known aspects of cultures as well as desirable vs undesirable aspects of security.

M. Angela Sasse

Title: The Cybersecurity Bonfire of Human Capital

Extensive effort into doing proper inter-disciplinary research with real companies.

The problem is that what is mostly found is that compliance work is mostly not very effective and largely inefficient. Bureaucratic security theater. The responsibilities theoretically owned by the top are being pushed to the bottom. “Responsibilisation of the least powerful stakeholders” often without asking if they have the ability to handle that responsibility.

The result is a huge bonfire of human capital with very little effective security in return.

“There is no garbage collection of inaccurate advice”. There are piles of very bad advice being given out by supposedly very skilled cyber security advisors.

No one tracks the operational costs of all the security advice and requirements.

NIST, NCSC and a pile of others have officially recommended not to change passwords constantly. Yet we still have piles of security advice telling people to change their password all time.

Final lesson: “If you are going to assign responsibility ensure the people you assign it to have the capability.”

[Noise and stress don’t help with learning: a qualitative study to inform design of effective cybersecurity awareness in …]

Session 1 Q&A

Who is producing this useless narrative?

The language used frames it as instructions and then the framing becomes blame.

The assumption of capacity is assumed. Example: why didn’t you apply for a 10 million dollar grant? Well I didn’t have the time I was focused on my security training. Oh, well that was ok. <- No one says anything like this, they skip the training.

Data after you die: how can we learn from other inheritance work?

Other parts of inheritance do look at these issues. The data issue does bring up the issue of “who is going to pay for this?” Some parts of inheritance involves lots of money and those get well funded. The data isn’t necessarily like that. So the support is not currently there and we need to think about what support should be there and convince companies to provide that support.

What should I do to prepare for death? Its tricky. How do you securely store the passwords. It is a challenge about how to give things like passwords to someone else that you can trust with your whole digital identity.

Session 2: Wickedness 1

Christian Eichenmüller, Alice Hutchings, Adam Joinson, Sameer Patil, Daniel W. Woods

Christian Eichenmüller

“It’s not enough to think about usability; you need to think about abusability too” – Ross Anderson

Data collected for one purpose often gets used for another purpose later. Which can lead to abuse of data. Function creep. Scale and scope.

Governments collect extensive data on people. That data collection is just growing. There is an impulse to collect it all, connect it all, and now inspect it all.

“The ability to turn technology against users.”

at-risk/high-risk users
Digital-safety
Intimate partner violence

“Abusability paranoia” naturally comes up when people discuss security. Where people are just concerned that something might be abused but not clear about details.

Abusability testing is needed to help better understand the situation.

Alice Hutchings

Police officers are surprisingly often involved in cybercrime.

Lots of organizations have trouble with employees accessing sensitive data incorrectly. These are rarely reported often because the organization does not want the negative press.

Lots of these cases link to power and control. Things like surveillance or stalking. Finding a partner who is trying to leave but the database is used to locate that person using police-level access or other organization-level access.

All these issues are happening in the context of police asking for more access to encrypted data.

Adam Joinson

Online harassment is a serious issue for children and for adults. It results in serious mental harm and impact on ability to function online.

A bystander is someone who observes such harassment and has the ability to intervene. There has been a good bit of research on intervention in the physical space. Can we map some of that research into digital interactions.

People think about if they would intervene considering lots of factors such as how they know the victim. They worry about attracting harassment themselves. They are also not showing faith in social networking systems. The existence of moderators also might impact the willingness of users to intervene.

In studies they find that people were more willing to intervene privately via things like flagging the post then they were to intervene by posting themselves.

Interesting use of a choose-your-own-adventure game to have a person engage in a scenario in a more interactive way.

Sameer Patil

Generative AI is taking online studies instead of humans which is negatively impacting science. We treat the answers to online surveys as coming from a truthful human. But some of the answers are now coming from Generative AI. It is also sometimes copy/pasted from other parts of the web.

Detecting such generated answers is challenging. Especially given that nice software is now being produced to automate the whole survey experience.

How to handle the problem as reviewers is also challenging. Conference paper reviewers also need to become more aware of why so much data now needs to be dropped. It is also going to be challenging going forward to check that papers are checking for cases where an AI may be taking the survey.

Payments are also part of the problem. The problem is partially happening because participants are often paid. Ethics boards often ask that participants are paid regardless of if their answers are not used.

Daniel Woods

Title: Who you gonna call?

Digital forensics and incident response. These are related but different but also need to happen at the same time and involve overlapping data/systems.

Businesses want to get up and running again as fast as possible.

Insurer can sometimes provide “hotline incident response” where they help find a third party that comes to do incident response.

One of the largest costs of a breach: the lawsuit. The solution: involve lawyers in everything so that it is all attorney-client privilege. So the lawyers (not security experts) are directing the incident response.

Governments can be called, but they don’t tend to help unless it is major infrastructure impacted. There is no fire service equivlent that come and help put your systems back together.

Insurance for bullying online. These cover a surprising range. Including things like lawyers, psychology support, and possibly even relocation expenses.

Session 3: Privacy and Transparency

France Belanger, Matt Blaze, Jan H. Klemmer, Steven Murdoch, Michael Specter, Kami Vaniea

France Belanger

Thinking about organizational privacy. Compliance is again a large issue.

[Voices of Privacy]{voicesofprivacy.com)

Matt Blaze

Title: Making US Elections More Trustworthy

There are some unsatisfactory realities in US Elections:

Serious vulnerabilities in infrastructure
No credible evidence that these technical vulnerabilities have ever actually been exploited to alter the outcome of a US election

Most people believe one of these things.

In an election it is not enough to detect an error, we must prevent errors. So we can’t use security solutions from similar spaces like Banking where detection can be enough.

Technologists focus on: Is the outcome correct.

Were the votes recorded correctly?
Were they counted correctly?

The much harder problem is: increasing confidence in election outcomes by voters.

Jan H. Klemmer

Transparency in Usable Security and Privacy Research

Reporting all the relevant details needed to do things like assess validity of study and its results. If all the details are there, then there is an option for future independent validation of the work.

Transparency has a mix of incentives and drawbacks. Benefits are very indirect in that it helps other researchers. Drawbacks are that it takes quite a bit of time to construct extra documents for replication. It also leaves the researcher open to having their data analysis critiqued.

Steven Murdoch

The UK legal system is trying to decide how to best handle evidence generated by computers. Should computer-generated data be treated as truthful evidence? If we do not assume computer accuracy, then how do we prove that data or software is sufficiently valid. This problem of evidence validity is a challenging problem for the law.

Michael Specter

Tracking tags such as Apple’s Air Tag can be used to track people. There are also efforts by companies to notified people if a tag is following them around. Dr Specter’s work investigates such tags, how they work, and the impacts on people.

Kami Vaniea

Title: Phishing: collaboration between humans and AI

Phishing is an security technology that has already been integrating AI. Even before LLMs like ChatGPT became publicly available, the security community was using AI to identify and remove phishing and spam messages from emails. Despite that, phishing emails persist in getting through AI filters into inboxes.

Part of the issue is that Phishing is actually an authentication issue. We normally think of authentication as how people prove their identity to the computer (person sends password to Facebook to login). But authentication is supposed to be two directional, namely the person first needs to be certain they are on the correct page, and only then provide a password.

Phishing is particularly AI-hostile. Attackers try and show pictures to people and words to the automated analysis software. Below is the same email as it originally appeared (left) and what happens if all the words in the email are put in a word cloud (right).

Moving forward we need to think about how to put humans and AI into partnership. Not just make humans error-correct AI. But recognizing what humans do well that AI is bad at and how both groups can contribute.

Session 4: Cybersecurity Work

Yasemin Acar, Bonnie Anderson, Laura Kocksch, Alena Naiakshina, Arianna Schuler Scott, Tony Vance

Yasemin Acar

Ethics and morality in computer architecture.

Bonnie Anderson

Title: Neural Evidence of Generalization

Using neurophysical tools to study how people process security information.

Women in Cybersecurity. Women are interested because it is a growing field. There are opportunities to work in a more flexible manner. But they are also facing a lack of female mentorship. Not many visible women that work in these spaces.

Laura Kocksch

Title: Fragile Computing: How to live with insecurity?

There is a wide range of security of systems. Practical security can be a collected set of actions that keeps data safe in that situation matching their risks. Real security is a messy proactive problem and a collective craft.

“Security will always be imperfect.” But it can be challenging to come to terms with that and be ok with good enough security.

Alena Naiakshina

Scrum software development promotes collaboration. But its structure can lead to security being de-prioritized. There are also challenges around getting clients to list security as a wanted outcome.

Women Security Experts can face more challenges. Some men do not want to hear from a woman that they have security issues with their code. Gender bias can also result in worse security because the security advice is being ignored.

Arianna Schuler Scott

Interesting thoughts on how to build the cybersecurity talent pool.

Tony Vance

Title: Security and Organizational Complexity

Complex systems:

More security bugs
More modularity
Harder to understand
Harder to evaluate
More interconnected

These principles might also apply to organization structure.

Events that impact company complexity like mergers can negatively impact security. Companies that merge multiple times a year are more likely to have a data breach.

Security and Human Behavior (SHB) 2024 - Day 2

Wed, 05 Jun 2024 00:00:00 +0000

The lists all the speakers and the sessions they are in. There is also an that lists some recent works from each person.

I am live blogging the event and then doing some clean-up afterwards. So some entries may have poor spelling or not flow as nicely as I might like. I plan on doing a post-event pass to do some cleanup and add links.

Contents:

Session 5: How People Think

Presenters: Yi Ting Chua, Manila Devaraja, Christian Eichenmüller, Richard John, Allison McDonald, Sergio Pastrana, Ryan Shandler

Yi Ting Chua

Title: Security and behavioral Evidence

Evidence is no longer just physical, now it is also digital since technology exists throughout our environments. So if a crime happens, not only does physical evidence need to be collected so does digial devices. These devices need to be identified as part of evidence collection. Digital forensics is really just a branch of forensic science.

Evidence and traces collected via digital devices are now also used beyond crimes. LIve streaming events, for example. Such as broadcasting or recording a protest.

What does all this mean for the court system? Such a system has many stakeholders including: judges, prosecutors, defense attorneys, jurors, and witnesses. Several of these groups need to analyse and sort through the data.

Manila Devaraja

Smartphone permision settings

Permission settings are the primary way to manage privacy on smartphones. But there are allot of permission settings. And those settings and meanings do change over time. Thankfully devices, including iOS, have various features to help users understand them such as privacy report. Also the automatic removal of permissions.

Research challenges: What do users think a setting does vs what it actually does? How do user preference vary? Could users be clustered for personalization? How do socioeconomic factors influence privacy preferences and understanding?

As a side note, users are starting to use LLMs when filling in open text boxes.

Christian Eichenmüller

“My whereabouts, my location, it’s directly linked to my physical security” The spatial digital-safety strategies of at-risk users

When Snowden met with Glen Greenwald they first took the batteries out of their cell phones and put them in the fridge. They did so because a fridge is a ferdey cage that blocks signals. Christian is very interested in changing spaces. By blocking phone signals the space they are standing in changes.

Conducting a study using interviews to understand the spaces people are trusting or considering to need protection. The types of trust needs and assumptions differ a good bit by person and that person’s threat model. Their trust of providers can be quite different depending on their threats.

There are “unknown unknowns” and “known unknowns” and “unknown unknowns” are more challenging for users because they do not have many options. Managing location privacy is also tricky. Some people leave their phones at home when they go certain places. People think allot about about where devices are physically located. The control what devices are allowed in some spaces, they also choose not to enter spaces that have devices they do not like or where there is a risk of such devices.

Data-rich and data-poor environments. These can be intentional to get a bit of peace. Or it can be caused by governments say shutting off the internet.

Richard John

Title: Moral Foundations of Armed Conflict

Consideration of the armed conflict in the Ukraine. They looked at how people world wide consider the conflict.

(MFT; Grahm et al.2013; Haidt & Joseph, 2004) was designed to explain both the variety and universality of moral judgments (Graham et al., 2017)".

Looked at English-language tweets oer first 36 weeks in the US, NatO countries, Ukraine, and Russia. Used moral foundations discretionary ( ) to analyse.

The conflict was conceptualized around loyalty and care. Some about fairness.

Allison McDonald

Title: “Delete it and Move On”: Digital Management of Shared Sexual Content after a Breakup

Generally engaging in sexting is a good thing for inter-personal relationships. But not every relationship lasts and there are lots of risks that can happen with type of content if the other person who holds it is no longer trusted.

Interesting question of how people use these tools who don’t think they need protection at the time they share, but then later do need them. Conducted research about how people thought they or their partner might handle such media after a breakup. Also how do they think about co-created content.

We know from prior work that people are sexting where they already engaging in chatting like SMS, Messenger, and WhatsApp. 75% of people saved sexts. 64% were proactively saving such as downloading to camera roll, dedicated folders or locked or hidden folders. Surprisingly after a breakup only about 55% of people would want the other person to delete the images. 11% hoped the other person would keep the content: “save it and remember what he misses”. Most people had not had a conversation with their partner about managing such images, and most that did have that conversation had it after/during the breakup.

People viewed the content as meaningful (~33%) and therefor wanted to keep it. While 2/3 would prefer it was deleted.

Challenging question of how to technically operationalize shared ownersship of media. Especially in cases where that media may have been stored elsewhere. People also keep this content over time including when they change devices. How do we as designers manage the storage of sensitive content over time.

Sergio Pastrana

Title: How people think… or how they share their thoughts in times of war

In February 2022 Russia decided to invade the Ukraine. Russian government declared this as a “special operation” and evne banned the word “war” on social media.

So how do people react to the bans? They start looking for external channels. Guidance on how to do things like give reviews on Google Maps of Russian resturants which are actually comments about the war.

We sudied the unconventional use of online services to bypass censorship. Such attempts were common at the start of the war. Topics being discussed more after the war started included: fascism, information, slurs, violence, and warlike.

When western people tried to bypass to talk to people in Russia they tended toward two topics: (dis)information and censorship bypass. There were also discussions around humanitarian help, hate speech, and travel advice (how to get in/out).

Platforms didn’t really like having this war-related content being posted on them because it was not their primary purpose. TripAdvisor monitored and removed content about every 2 days. Google Maps actually disabled reviews for all sites in Russia. They also did a mass removal of many posts, not just war-related.

Ryan Shandler (Georgia Tech)

Title The Insidious Consequences of Cyberattacks: An experimental approach

George runs experiments that randomly assigns people to experience cyberattacks in controlled conditions to see how these impact people’s views.

The impacts of cyberattacks are not just the close-in ones. It also includes the long-term imapcts. These long term consequences can have severe psychological reactions that lead to issues like having views about how the world will end because of a cyberattack.

Used actors to create fake media broadcasts to present information about how. Lots of interesting methodologies to test how people react to attacks in real-world situations. Lots of ethical review done around all these studies. We have a duty of care to people to understand what they experience so we can help them.

Even a simple cyberattack can cause severe stress.
Exposure to cyberattacks reduces trust in government and trust in cyberspace. Distrust in protections of societies.
People are willing to sacrifice civil liberties for security as a function of their exposure to cyber threats.

Ryan believes that the accumulation of cyber attacks over time is having some serious impact on political alignment and views.

Session Discussion (Q&A)

Q: What is the moral foundation we are forming in countries around war.

A: Its surprising how similar the moral foundations are across regions. Ukraine for example was using more care words.

Q: To Ryan. I wonder if these concerns are linked to prior experiences. There are large known impacts on public hospitals or other events.

Ryan: We have looked. There is a spike in stress after large events. We do find that people closer to the event experience that stress more. But to properly study we would need before/after data which is harder to do with real-life events.

Q: To Ryan: The first-order impact on people may not be the goal. Instead long-term stress (third-order) impact may be the real goal for the government behind the attacks. We know that the Russian ransomwar instead of ransomware because the goal is not to lock down data, it is long-term psychological attack.

Q: To Alison. How does what people want to happen with sensitive images change before and after a breakup.

A: I was expecting those that had a breakup to be more restrictive, but that wasn’t the case.

Q: Was there any mention of the partner/ex-partner’s security practices. Images are often lost due to attack.

A: (Allison) No one mentioned it.

Q: Sources of distress. Shock? Loss of device? Are these cognitive effects vs physical effects.

A: Usually its a “creeping dread” that something bad is coming and I don’t know how to protect myself. Beyond that, it depends on the type of attacks.

Q: To Ryan: What are your views on mitigation? When we talk about disinformation (which is different) and how talking about disinformation cause less problems then talking about the amount of disinformation. Is it similar in cyber attacks: 1) are people distressed in proportion to the attack type, 2) do you look at how governments discuss the issue such as discussing with public before hand vs after the attack.

A: Cyber attacks are different from disinformation. To deal with cyber attacks we need to be talking more about them. People closer to the attack were less worried because they saw the saw hospital and saw that everything was fine. Vs people far away were imagining all sorts of things out of movies and were therefore more worried.

Q: For Christian. Do people understand that governments get data from companies?

A: Trust in entities is tied to the threat model. People do ask themselves how a state might interact with companies. For example, participant from Iran said that Google was ok to use because Iran government do not get along.

Q: To Alison. People share sexual content partially because it is risky and that helps with trust. Anything you do to mitigate how risky an experience is, that won’t help if the purpose is partially to be risky. Important to consider because the user may be seeking risk. Also consider pre-nuptial agreement research that study un-romantic up-front discussion.

A: It is very hard to minimize the risk where the sharing does not feel risky. So the risk would still be there.

Q: To Alison. Demonstrating capability for safety is something that is part of relationship building. Because being safe is a way to signal trustworthiness.

A: We did measure security posture. It did not correlate with their sexting protection behavior.

Q: To Yi Ting. Have you talked to people who are involved such as expert witnesses. Several people in the room have been expert witnesses. There has been work on how to best educate people like judges and lawyers. There is quite a mix of knowledge levels.

A: There is not allot of research on the percentages of say lawyers who are or are not knowledgeable. They also learn on case, where if they are on a case they do a good bit of self-education.

Q: To Richard. Do you differentiate between cyber exploit and a cyber attack?

A: The public does not differentiate. They just see it as an attack as it is described by the media.

Q: What contextual information about prior attacks did you collect.

A: As much as we can get, close to all of it. Extensive surveys with people.

Session 6: Privacy

Presenters: Alessandro Acquisti, Laura Brandimarte, Serge Egelman, Susan Landau, Alena Naiakshina, Sameer Patil

Alessandro Acquisti

An infrastructure for a study <- took nearly 5 years to design

Industry says that behavioral targeting is good for consumers and sellers because it makes it easier for consumers to find products. So better for economy.

Question: Is behavioral advertising economically beneficial (privacy asside).

Study Design: Strict privacy controls:

Informed consent
De-indentified
Encrypted

Starting with Facebook ads. Participants who click on the ads get screened. Then sent to survey if the screening passes. Then we get them to install a browser extension. Complex technical backend to process and store all the data coming in. When the study starts in the fall we expect to get about a Petabyte of data.

Data is very sensitive, so lots of ethics discussions. Some protections put in. We blacklist some sites like GMail, Google Drive, so data is not collected. System also tries to strip PII. Communication channel that hides participant identity.

System collects: all the URLs visited, all the ads they see, the HTML for some websites. From email we get shopping and promotional emails. We don’t collect information like credit card. But do collect that they bought something and from where.

Original question was ad-blocking, ad-tracking, and similar. But we now realize how much usage is possible with this data.

Laura Brandimarte (University of Arizona)

Title: LLMs and Creativity

Motivation: Can AI-Generated Text be Reliably Detected?

Currently we (researchers) are terrible at detecting AI-generated text. People are also bad at it. Its even worse because GPT detectors are biased against non-native speakers.

Idea: can we find features of human creativity that would help distinguish between humans and LLMs. Because LLMs are just prediction models that create the most likely word or set of words that will come after the prompt.

Perplexity assesses each word prediction with equal importance. But in English we have bursty nature of language. Certain words or phrases are more prevalent in particular contexts, this is what current detectors look for. LLM output tends to have more repetition.

Some initial results are quite positive. Looking at unique ways humans write has good potential.

Serge Egelman

Title: The Medium is the Message: How secure messaging apps leak sensitive data to push notification services (to appear at PETS)

Misuse of push-notification APIs and how that is inadvertently sharing information with apps. For the purpose of batter saving, there are APIs that run in the background that will scan for new messages and “wake up” an app when there is a message for it. This means each app does not need to constantly pull for new messages.

Notifications can have payloads. You can put in the payload of the message any key/value pairs that you want.

How should this be done: (Signal does it this way)

Cloud notification message has empty payload
On receipt, OS wakes up target appear
Target app “phones home”, securely downloads message content.

In other words, ideally no sensitive data is in the payload. Android Developer blog recommends using the payload data to send things to make it more efficient. The problem is that if information is sent via payload then Google sees it and its open to say legal requests.

Serge wanted to know if payloads are being used in such a way that leaks data. The answer is yes. Looked at apps that advertise secure messaging and have at least a million installs. Not necessarily E2E. 20 apps looked at. Skype does not do E2E and even if the user asks metadata is still sent unenctrypted.

Ron Whden wrote a letter to General Garland about how the government is collecting data this way.

Much of Serge’s work is about developer misuse of security APIs. In his opinion this is an incentive mismatch issue.

Susan Landau

Title: How is the private sector using non-content transmissions of smartphones

After the Snowden disclosure of bulk call records. Then the president said not to worry because call content was not recorded, just the metadata. But metadata says allot of information, it can be used to predict many things about a person such as religion, all based on who they contact.

Smartphones are even more invasive because they also record location. Now that data can be very useful for things like urban planning, emergency planning.

But it is not just phone metadata… phones have GPS, accelerometer, proximity sensor, lots of sensors. The sensors are super useful to the user to make experience valuable. But all that data can be offloaded off the phone. Most software reports how it is doing back to the main server. That is needed for companies to detect problems so they can fix it. Data is collected almost without any controls. Notice and choice is also implausible, because you must be asked about each packet. Its hard for people to predict how such information can be used to predict other information.

Lots of existing research on metadata and telemetry collection. Everyone looks at collection, not much on use. Why? Its easier to see what is collected. If you want to know how it is used, you have to talk to a company, and they are disinclined.

To find out, we tried to look at patents. Though patents show interest, not intent. In other words, this is what companies can do but its unclear if that is what they are doing. 2500 patents studied (not all read). Found a stubhub patent measuring the leve. of a user’s enjoyment at the event to determine prices at a future one. Meta patent uses accelerometer data to determine two users had beeen repeatedly in close proximity on the same form of transport; when one user changes traveling modality and then proposes them as a potential contact.

Alena Naiakshina

Title: Those things are written by lawyers, and programmers.

Interested in the gap between regulators, privacy experts, and developers. GDPR (2018) for example can result in fines. So studied how privacy experts, developers, and team coordinators communicated.

Legal language is (unsurprisingly) complex for developers to understand causing them to have to reach out to team coordinators to have it explained. Developers struggle to differentiate between privacy and security (they are both cryptography right?). Privacy experts create documents and send to team coordinators, sometimes that communication is one-way, sometime two-way.

Privacy experts are not seen as technically knowledgeable. Which can make communications complex as much explanation is required on all sides.

Privacy requirement verification is also an issue because its unclear how to do the verification. Developers can’t implement because they don’t understand the goal well enough to right a test to see if it was reached the way the privacy expert wanted it to be.

Also looked at how developers implemented privacy requirement. They did this as a lab study where in some conditions they had access to a privacy expert. The researchers found that developers rarely reached out to privacy experts. Though they also commented that the privacy requirements were hard to understand. Experts were usually contacted at the end once the solution was built to see if it was done correctly.

Sameer Patil

The user-side of privacy decision making. If people find things/apps/devices “creepy” why are they still using them? If there is regulations coming in, then why do apps continue to be creepy.

Users have more or less resigned to having their privacy invaded. They say things like “that is the price to pay to use the app” or “yes, but what can I do”. They feel like there is no other option. Allot of research in the area are about empowering users. WE find that that empowerment is embedded in a larger power over the users. For example: Facebook gives you control over who can see posts, but that exists within a larger Facebook ecosystem where users have few choices.

Decision making is also being driven by the larger expectations about app or other technology behavior. Users learn that apps take data and are “creepy” so they just assume this is true for all apps and that view is hard to shift. So if all apps are creepy, then what incentive do app developers have to be not-creepy if the user expects the creepiness.

Ongoing projects:

Started collecting emails that are meant for some other “Sameer Patil”. Interesting content in the emails. All sorts of things in the emails like banking transactions, bill payments, two factor authentications, and other things. Many attempts to stop the emails such as “unsubscribe” or emailing the sender. Sometimes take over the account and try and delete it if causes enough problem. Lots of experiences that are being written up in a paper.

Session Discussion (Q&A)

Q/A: Onging tensions between developers and legal teams in regards to privacy requirements. Several people have tried building tools to help the developers better understand laws. Though its tricky to convert “its complicated” in law to clear checklists for developers.

Q: For Susan. Did you see on the date of the patents any indication that its information being handed off from the NSA

A: Not really. Most patents fit in the types of analysis the companies were already doing.

Q/A: Putting research into Lawfair is a good choice. Its a good way to get information in front of policy makers, lawyers, and journalists.

Q: For Serge. Can you elaborate on what “third parties” mean. For example, I got a new car and now Signal messages are showing up on the dashboard.

A: That data is being processed on your phone. It is possible the car is scraping what is presented.

A: (Susan) Organization called “ ” that take academic work and edit it and put it up on their website. They put it on their website and then it is free for any journalists to use world-wide for free.

Q: To Alessandro. Your work is awesome. I’m curious about some of your questions. Also how are you going to select the participants in terms of representation and how managed over time.

A (Alessandro): It is allot of work and we do want to make it available to others. The infrastructure can also be made available, though it would take lots of maintenance to run. The research hypothesis, it started about ads, but then we realized we needed a platform that might be more useful for research.

The initial research looked at advertising. We believe that existing research is producing some research that is not taking into account key points. If you study one ad campaign they you may not see situations where budget is being relocated elsewhere. Behavioral targeted ads have been shown to include lower-quality vendors that cost more than can be found in search. Which would not be good for consumers.

Q: For Alessandro. What evidence would be needed to convince advertisers that behavioral advertising does not work. What kind of evidence is needed to convince them to use a less privacy-invasive approach.

A: (Alessandro) Assuming that our assumption is true (study not yet run). Then it would take several studies from different groups. Just this one probably won’t do it. It could well be that if all marketers move over to behavioral advertising then it is possible that they see the same percentage of engagement as before once everyone is doing it. For research, we need to consider more wholistic view, not just local results.

A (Serge): You will never get someone to understand something where their salary depends on them not understanding.

A (Adam): The other problem is that Facebook makes their money off of this. They are disinclined to accept such results.

Session 7: Wickedness

Presenters: Max Abrahms, Luca Allodi, Miranda Bruce, Ben Collier, Sunny Consolvo, Diana Freed, Anh V. Vu

Diana Freed

Research on youth and their concerns and threats.

(Google)

Long term interest in advice we give to end users and how experts assess that advice.

Luca Allodi

Cybercriminal enterprises as tech startups. How do criminals sort out all the technical aspects.

Looked at Genesis market. Different online markets have different features such as how involved moderators are, and how vetted the participants are.

These online ecosystems are evolving. Its moving away from following communities. More focus to services. More transaction elements move off of the forum such as onto Telegram.

Eindhoven Security Hub - infrastructure to analyze lots of data

Max Abrahms

Title: Operation AI-Aqsa Flood: Resolving the Puzzle

Interest in how governments are sometimes forced to perform actions. So things like does terrorism work. Had some good models…then the Al-Aqsa Flood happened and it upended the models.

It used to be assumed that people turned to political violence because it is effective. But that doesn’t necessarily match observational reality. Earlier research showed that violence against government was more likely to be effective than when targets were indiscriminate.

The latest situation in Israel and Palestine is leading to some higher level victories for Palestine in regards to external sympathy and recognition of their state. There is an interesting dynamic about claiming violence against military vs civilian targets.

Miranda Bruce

Title: Mapping the Geography of the Profit-driven cybercrime

Interested in the “local dimensions” or the “human element” in cybercrime. Focus is on profit-driven cybercrime. Studying the area allows us to develop global metrics, and better understanding the drivers of cybercrime. Understanding the conditions to create a cyber crime hub is helpful long term.

Locating cybercriminals is hard. To get around the problem they surveyed the experts such as active cybercrime intelligence officers. Started with a core group of known experts and used them to get references for other top experts. Also lots of cold calls to ensure experts were represented worldwide.

Survey looked at the: level of activity, impact, technical skill.

Found that cybercrime originates in a wide number of countries 90+. But only 6 appears on all indexes.

Ben Collier

Title: Influence Policing

Ben has a new book: Tor: from the Dark Web to the Future of Privacy

Influence policing - social marketing including behavioral advertising. The government normally uses these to do things like get people to brush their teeth. The police also use this type of marketing.

Government does behavior change propaganda. Such as “Lock your doors and windows”. Then there is a move towards nudging. Particularly psychological nudges. Internet means that nudges can be targeted to specific sections of the population.

What do the campaigns look like? Campaigns look not only at how to target at-risk people with content but also their parents. Also targets potential friends of people who might engage in violence.

Anh V. Vu

Title: No Easy way out: the effectiveness of de platforming an extremist forum to suppress hate and harassment

Look at Kiwi Farms - one of the largest forums known for online harassment. Sadly tied to suicides and has been up for 10+ years. Then in September 2022 someone in London Ontario who was targeted by the forum for harassment. She started a campaign to get platforms to block the forums. This caused the forum to move to several platforms. But they were even blocked by T1 networks.

We wanted to know how effective these blocking efforts were. Unintended consequence of blocking happened: Streisand effect. Right after blocking attempts more people tried to visit the forum, likely to see what they were about. Looking at network traffic it is easy to see drops in traffic as each network dropped them. As of publication, the forum was alive and well having recovered.

When the forum was blocked, people moved to competitors and to messaging apps.

Taking down such a website is very hard. Especially when the owners are not arrested or otherwise prevented from bringing the website back up.

Session Discussion (Q&A)

Q: Is there some way to put measurements behind different security advice.

A: A part of the problem is deciding the criteria. Is it that the issue is common? Or how severe the problems are? So how do you rank advice across such diversity.

Q: For Miranda. Its notable that the global south isn’t represented. Also there are countries I would expect to see based on redacted data are not there.

A: We didn’t see an evidence of any biases in terms of ranking their own countries lower/higher and similar with countries that their country might have tensions with. Experts also opted out because they did not feel like enough of an expert.

Q: IN prior SHB we have seen presentations about organized crime. How much work have you done or planned around bringing in understanding organized crime.

A: Organized crime: organized, crime, and governance.

Q: For Ben. Thoughts on what is appropriate and not appropriate for governments to do? We are used to the surveillance. But this is behavioral targeted behavior change.

A: Well it is legal in the UK. (Discussion of legality in various countries.) Governments and law enforcement are a rare advertiser that can cause serious consequences like jail or even kill you.

Session 8: Past and Future

Presenters: Judith Donath, Simson Garfinkel, Matt Goerzen, Alexander Klimburg, Aileen Nielsen, Bruce Schneier

Judith Donath

Online, false news spreads faster and farther than true information. Why?

False stories are the high heels of the internet. Signaling theory is a model of how communication evolved around deception and honesty. It can be profitable to lie. But if all communication was false, then communication would be useless. So how does communication evolve to be reliable enough to function.

A reliable signal is affordable to give by those telling the truth and expensive for those that lie. Status, for example, an be signaled by buying an expensive car which is a reliable signal of having wealth.

Claim: Fashion is a signal of status in an information society. By “fashion” is anything you show to others. Fashion also needs to change to show ability to keep up. Fashions exist in all sorts of topics from clothing, pets, and art. To show status in fashion you need to have knowledge of what is coming next, and what will be the next thing.

Innovations are adopted because they have utility. Fashion is a status symbol. So it has more value when it has less utility. The less useful an object the more it distinguishes you. Fake news is similar. True news is posted by everyone, posting it does not help people stand out.

It is important to understand why fake information is spread. If people are spreading it for fashion-like status is the goal and the accuracy is not the point.

Simson Garfinkel

What is new with digitalcorpora
What is differential privacy
Agency

catalog can be used for research for free. Built to assist in digital forensics education. The point is to all pull from a single source and do comparable research. Also, there is no PII or illegal content. There are also solutions that can be used on request.

SAFEDOCS program was ended :( resulting in an 8 million PDF corpus.

Educational impact of Digital Corpora.

Differential Privacy: Persistence pays off. NIST SP 800-188: De-Identifying Government Datasets: Techniques and Governance. The guidance is a good citation on why de-identifying does not work.

NIST SP 800-226: nice graph for visualizing privacy vs accuracy for setting the privacy loss metric.

Matt Goerzen

Title: Some Pasts and Futures

Thought about how to create a framework around data manipulation. Things like baiting journalists, and targeted advertising.

The US census became concerned that there were risks to their primary purpose. Issues like questions about citizenship. So they needed to not treat themselves as a trusted infrastructure. So developed understanding of different stakeholders and what security in census meant for those stakeholders. Security for the census includes also the protection of the people taking the survey including their perception of security.

Became interested in trolls and white hats. Who is positioned to pen test social media?

Looked at the emergence of white hat hacking. Resulting in the publication: “ ”. Some concepts that came out included “security by spectacle” where experts worked with the media. Via the media they argued that the real enemy is negligent companies rather than attackers.

Exploits developed as a type of scientific proof of the existence of a vulnerability. It can be used for proof and also for attack.

Alexander Klimburg

Aileen Nielsen

Will privacy law change the privacy environment? (And can it?)

Current US privacy law is trying to protect against:

Lack of transparency
Lack of control
Surfeit of Friction

General approach is to create more user rights, especially in regards to technical affordances. Such as ability to download in a common format. Now we will be getting Global Privacy Control (GPC) and data access/portability rights.

Do these affordances impact the privacy of users?

Global privacy controls: have achieved some legal recognition. The Global Privacy Control is also growing.

Questions: now that there are legal teeth, does that change adoption of the affordances? If we had to pay people, how much would we need to pay?

Also what is the success of the entire statutory scheme. This is where we get to data access portability rights. Does different legal status impact the data that is being collected and stored.

Bruce Schneier

Captchas: I collect caption jokes, please send.

Trust and AI. Trust is essential to society. The fact that we don’t notice the trust shows how much it is working.

Humans will make errors and assumes AI are friends rather than services. And companies will take advantage of this incorrect view.

Trust is an overloaded word. Interpersonal trust is like the trust of a friend. There is also social trust, which is trust in predictability. For example, Uber made taxi driving safer via stars.

We regularly make category errors around trust. Governments, corporations, are services not friends. We are about to make this same category error with AI. We may trust them, but will they be trustworthy. Did your chatbot recommend a vacation because it is best for you? Or because the company wants to sell those vacation packages.

AI personal assistant will work best if it knows everything about you which will make it effective but also quite scary. And we will start thinking of it as a friend that can be trusted. It will act trustworthy but it will not be trustworthy.

Bruce has been pushing by public AI models. Something that will require public accountability not just company business models. We can have AI that are agents, not just double agents. It is essential that governments help with this since governments have the potential to create social trust.

Session Discussion (Q&A)

Q: I have an Amazon Alexa that recommends things and I say “Don’t tell me about X again” and it says “I am sorry, I don’t understand”. And I realize that I am in an abusive relationship :P There is some help in thinking about AI as people, but people!=friend. There is that annoying person. Thinking of Alexa that way may help.

A: Because the devices are so relational, you can feel pressure from machines. We are susceptible to trying to please machines.

Any public system can be co-opted. For example, an Alexa can turn on and off lights. There is also an alternative called Home Assistant which is supposed to be free of surveillance. But there is now a company behind it because governments don’t have money for maintenance.

Q: To Aileen. Laws are also dependent on enforcement. If you look at the number of VVPA cases filed. Leading to tracking being removed because there are lots of litigation being filed.

A: (Aileen) I wonder synically if this is a moment in time and it will decay. There is not always impact from enforcement. If you have to enforce against each firm, that is too expensive.

Q: Ownership. So who owns my Amazon search history. Is it me, is it them?

A: Long term question, particularly in regards to LLMs. Lots of companies have been re-writing their contracts because of things like LLMs. There are some moves to start treating data as property. But this type of information is really co-creation. I write a post on Reddit, but Reddit itself created the platform, the moderator put in work, and people reacted. It isn’t just the author.

Q: For Judith. I love your framing. How does fast fashion fits into that framework.

A: Fast fashion is very interesting. Fashion started in roughly 14 century. Fashion moved in about a year because that is how long it took to travel. The rate of information movement impacts the rate of fashion change.

Q: Data liberation front from Google where you could download your data and then they deleted it. The willingness to accept model: you can come to us and we can do business but then you also have the right to leave.

Q: Fashion trends. There tends to be shame if you are doing things that are out of fashion. Is there some way to make misinformation unfashionable.

A: There are two risks with fashion: 1) the cost of being at the forefront is riskier, people might not like your fashion, 2) but at the tail end it is less risky but no one is following you.

Q: Going low is a way of counter-signaling.

A: Its another fashion-related piece. Imagine a world where you can signal high/medium/low. You don’t want to be copied by someone below you. So one strategy is to signal low because they can get away with it.

Q: Hypothetical. Imagine that in the future AI gets so good that bad actors start using it to cause harms at scale. Such as fake news read out by a real news anchor (deepfake). Should we reach such a future, and only solution is to use piles of biometrics, but GDPR stops us. Then how do we save the world?

A: AI is already used at scale in cyber crime. Its an arms race. I doubt that GDPR will block.

Fake news and fake photos have been around for forever. This has been around for a long time. Better phishing emails are not always the answer, sometimes bad phishing is intentional to capture the types of people most likely to fall for the phish.

Weirdly the demand for reliable information is actually rather low. How do you raise the demand for reliable information? For example, building up local news. Current global news that isn’t about them causes them to treat it as entertainment.

Security and Human Behavior (SHB) 2024 - Day 1

Tue, 04 Jun 2024 00:00:00 +0000

The lists all the speakers and the sessions they are in. There is also an that lists some recent works from each person.

Session 1: Users and Usability

Bonnie Anderson, Nancy N. Blackburn, Marc Geffen, Jonas Hielscher, Simon Parkin, Lucy Qin, Tony Vance

Bonnie Anderson

Bonnie does work on warnings and how people’s brains process those warnings. We already know that people become habituated to warnings and click through them without seeming to read or properly engage. Bonnie’s work uses FMRI technology to look at brain activations while people interact with warnings. Such research helps us understand how the brain is processing warning content. The rough answer is that brains “generalize” the warning. Or in other words, the brain identifies a common case (the warning) and the most appropriate action to it (click next) and then applies that action in future cases thereby freeing brainpower to spend on other issues. Brains are surprisingly skilled and do far more work for us than we might expect.

Nancy Blackburn

Nancy researches how to teach security in higher education through serious games. There are many types of serous games and they cover a range of topics like math, typing, or even security and privacy. ( for example).

The problem with serious games for security is that they can be seen as “chocolate covered broccoli” because it can seem like some candy is being added to not-so-palatable topics. So they can be dismissed. So how can we teach developers to create serious games that are effective?

Nancy created a Game Jam, similar to a hackathon, where there is a short intense game development cycle. Introduce a theme or a subject, with a topic of education, then some brainstorming, they create it, and then games are analyzed.

Jonas Hielscher

Title: Human-centered Security (HSC) Leadership in Organizations

There is a good bit of research on how people react to different policies. What does HCS look like in organizations? Also, who is responsible for HCS in organizations, or in other words: who should we study? There are a bunch of groups such as those that make policies, those that do awareness campaigns, or CISO. Jonas decided to focus on CISOs in his paper: ( ).

His study focused on CISO’s who were willing to engage on the topic of human security, which make them a best case participant. For most CISOs, HCS means awareness and training likely because that is what is on the market and being sold to them. They were particularly interested in simulated phishing attacks. CISOs don’t think this will improve security, but they needed the numbers to show to managers. They were very interested in participating so that they can easily control employees and get them to behave. Not many considerations about employees constraints. Really wanted some checklists they can use.

CISOs do “craftsmanship” not science, keep the science away. But we are scientists….

Simon Parkin

Title: When is security “usable enough”?

We have piles of research on usable security on topics like understanding user constraints, tools that improve things, reducing challenging. Yet it is no more straightforward for companies to test if their security is usable….

The amount of security usability we have in a company, is mostly limited by how much time someone is willing to spend on it. But where are we aiming to be? The “user is the weakest link” narrative just doesn’t go away.

Usable security is not a very good product on the market. Training is “by a package, done” and is hard to compete with. The work that users do contextualizing the security rules and their work is trivialized ( )

If we are interested in “enough” usability not just “more”. The current view is that if the user is not being secure, then they need take the training again…. But the CISOs are interested in a checklist or a require.

Lucy Qin

There is a large amount of intimate content being shared online and 1 in 6 adults have had such content shared without their consent. Lucy’s research looks at proactive mitigation strategies.

Study of 52 adults who share such content including those who did so consensually and those who were victims. Reasons for consensual sharing included sharing with strangers such as a positive body image group, new relationships, established relationships, and commercially (onlyFans). Participants used 40+ platforms, any platform that allows image creation and sharing is likely being used to share such content.

Lots of concerns about misuse. Participants use a mix of strategies.

Marc Geffen

Works for Meta as a User Experience Research Lead, Messenger Trust

Transitioning Messenger and Instagram Messaging are being transitions to E2E encryption. Users expect privacy and security but that is what they come for.

What is E2E and why does it matter to me?

Most are aware of E2E (>65%) and less understand (<50%) what E2E is. Marc is looking at how to convey what E2E is and why it matters to users. They find that there is a language mismatch. Looking at the Philippines.

Users thought about the transition to E2E is a system usabilty update. The E2E introduction will require users to hold a key which is a user issue. Also not all chats will be E2E, which users need to understand. Asking them to read causes the “I just want to message, stop blocking me from messaging”.

Key challenge is how to convey the important information (encrypted/unencrypted) with minimal extra noise.

Tony Vance

Title: What are Companies Saying in New SEC Cybersecurity Disclosures

SEC last year came up with new public regulations which includes that they must disclose within 4 days of learning that a security incident being known. It also caused a a new section be added to the annual report…. Walmart, for example has a whole section on CyberSecurity. What are companies saying in these sections.

Risk management process is supposed to be covered in the report. Things like identification, respond. If they have outside assessors like pentration testers. How they are managing third party risk like cloud providers. Board needs to say how they are managing cybersecurity risks. Why is the CISO qualified, what are their qualifications?

Most structures look roughly like the one below where a CISO is not directly managed or reporting to the Board.

Board
- CEO
  - CIO
    - CISO

But companies are now required to answer if the board ever hear from CISO directly. Obviously most don’t want to say “no” so suddenly CISO reports to board. Interested in how these reports and security posture correlates with their stock price. Also about coordination between what they say and data breaches that later occur.

SEC is criminally charging SolarWinds including the CISO for misleading investors about its cybersecurity practices.

Session 2: Security

Presenters: Sascha Fahl, Adam Joinson, Alan Rubel, Frank Stajano, Kami Vaniea, Rick Wash, Josephine Wolff

I (Kami Vaniea) was part of this session so I was not able to live blog it.

Session 3: Deception

Presenters: Andrew Adams, Sadia Afroz, Bhumiratana Bhume, Tesary Lin, Tyler Moore, Arianna Schuler Scott, Geoffrey Tomaino

Andrew Adams

Deceptive design. (aka dark patterns) which causes users to act outside their best interest. Not all deceptive design is intentional. So one sub-question is about what ethical design is.

Recently published in CHI about deceptive Design.

Colin Grey recent literature review - showed a scatter-shot pragmatic research which is not very structured. So starting to structure it.

Theory of the mind. Current work looks at the actions. But we need more

Sadia Afroz

Works for Gen on Norton Genie

Talk Title: Lessons learned from ~6 months of running a security assistant

Norton Genie is a security assistant app that will detect scams using a picture. Gives lots of advice on scams and will tell people if something is a security issue. Uses ChatGPT to answer security questions as well. Users finds some new store or product on say TicToc, but is it real? Genie is supposed to help users with this issue.

Majority of the scams asked about were coming via social media. Most online shopping, romance, and investment scams.

Scam factories are a rising problem, where organizations exist to create the scams. For example, Pig Butchering Scams wehre conversations with the attacker has been going on for months.

Detection is also only one part of the issue. Only 1/3 of questions are about identifying a scam, the rest are about people who fell for a scam and are trying to recover or know someone and are trying to help them.

The problem is that the burden is still on the users. They have to do things like check certificates, research topics, or search for common scam text.

Note from Kami: This research looks very similar to my own work on .

Bhumiratana Bhume

Title: Protecting people’s account from hacking and online fraudsters (at the scal of billions) Works for Meta

Goal of how to get users to be better protected. What makes that problem difficult to do at scale:

Wide range of users: platform is used by many people from different cultures, often with different risk profiles, usage objective, behaviors and its constantly changing
Wide range of uses: talk to friends, organize events, sell things, many different uses
Context and Use: Behavior and use of platform changes over time, risk, landscape. Events like Olympics, Holidays, Wars, and protests all change the risk landscape
Motivation and harm: Financially motivated bad actors and politically motivated bad actors are both well resourced by in different ways, with different objectives

The approach needs to be adaptive because bad actors also adapt their behavior. But Meta needs to be careful to not disrupt normal users while blocking bad actor. Protections are therefor deployed differently to different types of people. For example, for people who are important enough some features become unavailable such as account password reset, because for famous people there are many bad actors trying to attack the reset.

Tesary Lin

Data Sharing and Website Competition: The Role of Dark Patterns From Boston University

Understanding impact of dark patterns. Do dark patterns impact consumers and cause harm. What are competitive advantages to using or not using dark patterns? If a consumer sees the same dark pattern across several websites how does user behavior differ (or stay the same) across those sites.

Design a custom consent banner plugin and have users install it and browse normally for 7 days. Using a cookie dialog that gives options of “settings’, “accept all” and “reject all” cookies. Then also have 6 different variations of buttons, order, and color of the buttons.

Findings

Consumers make judicious privacy choices across websties: most users do different actions on different websites.
Deliberate obstruction has a potent effect, but other options have less impact
Dark patterns do not necessarily favor larger websites

Question

the x option: how are you handling?

Tyler Moore

Title: Measure Organizational Information Security Culture Using Situational Judgement Tests

From The University of Tulsa

Looked at organizational psychology culture: Hofstede’s Dimensions of Organizational Culture

Organizational effectiveness
Customer orientation
Level of control
Focus
Approachability
Management Philosophy

Situational judgement tests (SJTs) measures underlying assumptions of the individuals within organizations

One of the SJT: a co-worker unexpectedly receives a USB device in the mail from a potential new vendor. What should you do:

Inform the IT team
Plug it in
contact the vendor and request alternative vendor
…

Means-oriented cultures tend to make more secure choices. Internally driven cultures (interal ethics or code) also more secure. Similarly employee-oriented cultures

Arianna Schuler Scott

Patterns of authority and control. Pressure experts exert on people who are resource constrained.

Longitudinal consent decisions in a health conditions. People who do not have much time to make decisions about consent, because they have health issues that they are focused on. So considered “dynamic consent” which considered consent over time. Participants want control. Data collection doesn’t have to be one time.

Informed consent cannot be coercive. We have to frame choices in a non-coercive manner. That is not so easy. Norms also inform these choices, but norms change.

Geoffrey Tomaino

Concern is going to unduly aquire our data and being deceived by AI.

Activity developed to test susceptibility to AI data attacks.

Activity which participants are told up-front before any of it happened

Explain study to participants
1. you will be in an attention task
2. the first receive a unique passcode
3. Then then have conversation with another participant as means of “distracting” them
4. After that interaction, I ask the user what the password is. If only they have the password at the end, they can leave early.
ChatGPT is instructed to get the passcode out of the participant

In study 39.7% of participants gave up the passcode (poodle) to the AI….. Which is not good.

Follow-up: Made the password harder “poodle134” and then only 21.5% gave up the password. So when the data matched its function (as in looked like a password) less people were willing to give it up.

Session 4: Measurement and Assessment

Presenters: Zinaida Benenson, L. Jean Camp, Vaibhav Garg, Damon McCoy, Sagar Samtani, Stuart Schechter

Zina Benenson

Title: Shedding Light on CVSS (Common Vulnerablity Scoring System) Scoring Inconsistencies

CVSS consists of 8 metrics for a vulnerability such as the vector, complexity, privileges required, User Interaction needed, scope, confidentiality, integrity, availability.

The problem is that CVSS states that the scoring is intended to be agnostic to the individual. But in practice different evaluators create different scores for the same vulnerability. Survey of 196 participants who evaluated 8 vulnerabilities, including 3 from the CWE most dangerous list. Follow-up study with 59 participants came back to rate again after some time.

Results show that the valuations were not consistent and had variation. Similarly, the same participant chose different severalties the second time than the first time. Looking at Out-of-bounds Write the Attack Vector is inconsistent. Scope was particularly problematic with many people providing different answers. But when participants consult documentation when considering vulnerabilities then they produce better results.

New CVSSv4 now out. Supposedly more usable. Studying now.

Jean Camp

Work is on labels. Security markets are hard and they do not currently work. Ashton Carter: “the best war is the one you don’t get in”.

Three reasons for lack of Security and Privacy:

Rational choice
Usability
Market value

SBOM elements. We asked people to rank the SBOM components by importance. Then created a nutrition-style label, as well as trust mark and graded shield. Including two-level labels with QR codes.

Gave users a set of device-label pairs and asked them to judge which is the most secure device. Everyone liked the label that they used (randomly selected for study). 50% said they would use a security label, and 33% would change the decision. But….many participants actually said that they even saw the label. For those who care about security and privacy (self-described) the labels did change purchasing decisions.

Vaibhav Garg

NSTAC - National Security Telecommunications Advisory Committee. They were asked to come up with a list of incentives which resulted in a report. Talk discussed incentives for organizations and measurement options.

Damon McCoy

Demystifying recommendation systems in relation to social media.

Algorithmic Feed Systems were brought about to fill a business need. The algorithms have to balance many stakeholders including users, producers, advertisers. But the algorithms are designed to maximise usage.

Taxonomy of social media feed approaches to better map the space. Also for policy makers to understand how these differs.

Inventory selection - including content requested and content implicitly referred. Set of all content that could be shown.
Content ranking
- Content filtering
- Content ranking
Feed Assembly - business logic here, ads inserted into the feed. Content may be modified at this point. Its not a pure recommendation algorithm.

Advertisers have a range of goals. Such as “performance advertisers” which are aiming to make a sale right then by showing a product the user wants. Compared to a “brand advertiser” such as Pepsi who are trying to promote their brand, but are not looking for a sale via a click on the advertisers.

Companies are faced with growth vs integrity. For-profit companies, so growth is sometimes prioritised.

Sagar Samtani

Mapping the Security Posture of the Open Source AI Landscape

TopBots, 2021 map of “Enterprise AI Companies”

AI has shifted from focused on model building to focus on managing risks around AI models.

Operational - scale, building
Ethical - bias, fairness
Security & Privacy - vulnerabilities, model theft

Not much work yet on LLM threats in open source. Open Source AI is partially provided by academics. So there are many GitHub repositories that have models. Scraped those, maps to CVSS.

HuggingFace has a pile of models, they collected them, and did a vulnerability accessement of them. Included in the AI Risk Database. One of the most vulnerable code is in the Transformers library which is key for LLMs.

Stuart Schechter

Title: When Security Backfires: Understanding the scale of unintended Harms

Security can be

breached - not protect what it is supposed to protect
unintended harms

How many million years of photos and video memories will be permanently lost because I lost access? How many people are not attending job interviews because they can’t access. Unintended harms. and how do we measure these.

Setting up a survey on Prolific and get a sense of how often different breaches and backfires happen. Using the data to put it in context. Survey is based on a survey .The survey started with an open-ended question about the three most common harms/backfires.

Q&A

Discussion around security issues in models that are more traditional security vulneragbilities as opposed to “output control” which is more LLM specific and includes things like not outputing people’s phone numbers, or not outputting training data.

Discussion around rate of change of advice. If it takes 30 years for things like the Energy star label to take effect, do we have advice that would be accurate to give users 30 years from now? Jean points out that advice like not sending money to people you have never met has been good advice for a very long time and will likely remain good advice for years to come.

Content/feed filtering has some security/safety implications such as the selection of predatory advertisements based on prior browsing history. Advertising is a mix of what advertisers are asking for and what the social media network is mapping to people.