Starting in ~15 mins!

Improving Peer Review in Top-Tier Cybersecurity Conferences -- A proposal in 10 points 1. Introduction The peer review process at top-tier cybersecurity conferences faces well-documented challenges: inconsistent review quality, overburdened program committees, scope creep in submission ...

Check out our latest article: Improving Peer Review in Top-Tier Cybersecurity Conferences -- A proposal in 10 points linkedin.com/pulse/improv... via LinkedIn

Trustworthy AI Seminar Series Trustworthy AI Seminar Series

New exciting Trustworthy AI Seminar this Friday!

Earlence Fernandes (UC San Diego) will present "The systems security approach to AI agents."

Details and Zoom/YouTube stream link on our website : trust-ai-sem.github.io

Can someone explain to me why car dealers agents still exist? If there is profession at all one should be happy to see replaced by AI, car dealers are light years ahead of anything else

I'm excited to teach a Cybersafety class this quarter at UCR. In this first iteration, it'll be offered as a seminar class, but we will switch to a regular class next year.

Preliminary syllabus: spalab.cs.ucr.edu/teaching/26s...

Looking forward to comments and feedback!

Trustworthy AI Seminar Series Trustworthy AI Seminar Series

🎙️ New Trustworthy AI Seminar

⏰ April 3 | 9am PDT / 12pm EDT / 6pm CEST

🗣️Sahar Abdelnabi ( @sahar-abdelnabi.bsky.social): "Stateless Yet Not Forgetful: Implicit Memory as a Hidden Channel in LLMs"

LLMs can carry hidden state across sessions — with zero external memory!

👉 trust-ai-sem.github.io

Trustworthy AI Seminar Series Trustworthy AI Seminar Series

📣 The first seminar of the Trustworthy AI Seminars will be given by Peter Kairouz on "Context Matters: A New Paradigm for Differentially Private Synthetic Text Generation."

🕰️ Friday, March 6, at 9am PT / 12pm ET / 5pm UK time

🌎 Details, Zoom link, YouTube stream, etc. at trust-ai-sem.github.io

Hey everyone, @ebagdasa.bsky.social and I are rebooting and widening the old prisec-ml seminar series. Give us a follow at @trust-ai-sem.bsky.social and check out our website at: trust-ai-sem.github.io

We already have a couple of amazing speakers booked!

Doesn’t really instill a lot of confidence in the process :) but happy to help with anything privacy related

I’m excited to share that the UCR's LOGOS Interest Group is hosting an Open Day on February 27.

If you’re a Southern California–based researcher and you’re interested in research problems in cybesafety/computation social science, get in touch!

👉 Event page: logos.cs.ucr.edu/open-day.html

MAGA level copium!

Unapologetically wouldn't consider living anywhere in the US other than California.

You can get here with a 16 min train ride from Union station and a 10 min uber

PSA if you’re ever in the LA area, don’t miss Raffis place in Glendale. Worth the detour just for their khoobideh. Wow.

Just when you didn’t think this week could get better!

UC, 2024: Your health insurance is going up significantly because we cover GLP-1, so go and hate people on it

UC 2025: We are no longer covering GLP-1 unless your BMI is above 40 (if you're 5'10", you'd need to weigh >280 lbs)

Also UC, 2025: Your health insurance is going up because GLP-1

Wut?

To Shuffle or not to Shuffle: Auditing DP-SGD with Shuffling The Differentially Private Stochastic Gradient Descent (DP-SGD) algorithm allows the training of machine learning (ML) models with formal Differential Privacy (DP) guarantees. Since DP-SGD processes t...

Congrats to Sundar -- his paper "To Shuffle or not to Shuffle: Auditing DP-SGD with Shuffling" was just accepted to #NDSS2026!

Only 21 out of 950 straight accepts this round 😳

Joint work with Borja Balle, Jamie Hayes, & @emilianodc.com

Pre-print: arxiv.org/abs/2411.10614

Hey folks! My Fall quarter class (OS capstone project) has been canceled, which means I'm on an impromptu short sabbatical until Xmas.

I'm happy to travel a bit to visit and give talks if anyone is interested!

I'm aware that I'm known as a hater, but can someone explain to me why on earth anyone would willingly use Overleaf to work on papers?

The only use cases that I can think of are:
1) emergency editing from someone else's device or iPad
2) non-CS co-authors not familiar with LaTeX

You can keep in and out. Californians are happy to get rid of them.

Ever struggled with managing your pubs to create an NSF BioSketch using MyNCBI's My Bibliography? I think I have finally found a decent way to handle stuff, sharing in case it helps... (Step 5 is the one that took me the longest!)

www.linkedin.com/posts/activi...

The Importance of Being Discrete: Measuring the Impact of Discretization in End-to-End Differentially Private Synthetic Data Differentially Private (DP) generative marginal models are often used in the wild to release synthetic tabular datasets in lieu of sensitive data while providing formal privacy guarantees. These model...

Work led by my student Georgi Ganev, with Meenatchi Sundaram Muthu Selva Annamalai and Sofiane Mahiou. See you in Taiwan!

arxiv.org/abs/2504.06923

4/ We introduce a new bin selection method and demonstrate that DP-aware domain extraction significantly reduces privacy leakage

5/ We once again stress the importance of privatizing preprocessing in DP pipelines.

Highlights:

1/ We evaluate four discretization strategies across different models and datasets
2/ Optimizing the choice of discretizer and number of bins can improve utility by almost 50%
3/ Failing to privatize the discretization step can severely compromise privacy.

📣 Upcoming CCS'25 paper 📣

The Importance of Being Discrete: Measuring the Impact of Discretization in End-to-End Differentially Private Synthetic Data

We investigate the critical yet overlooked role of discretization in end-to-end differentially private (DP) tabular synthetic data generation

The Importance of Being Discrete: Measuring the Impact of Discretization in End-to-End Differentially Private Synthetic Data Differentially Private (DP) generative marginal models are often used in the wild to release synthetic tabular datasets in lieu of sensitive data while providing formal privacy guarantees. These model...

Conditional congrats to Georgi and Sundar - their paper on Discretization in DP Synthetic Data was accepted with shepherding to CCS’25.

Very important work on an overlooked aspect of end-to-end DP pipelines.

Preprint: arxiv.org/abs/2504.06923 -- feedback welcome!

Bonus highlight: In at least one case, a broker leaked another person’s sensitive info, including a driver’s license number.

Consumer Beware! Exploring Data Brokers' CCPA Compliance Data brokers collect and sell the personal information of millions of individuals, often without their knowledge or consent. The California Consumer Privacy Act (CCPA) grants consumers the legal right...

Really interesting work on CCPA-mandated compliance of data brokers with access/deletion requests: arxiv.org/abs/2506.21914

Highlights:
- Almost half of the brokers don't respond
- Only <5% of brokers provided personal data
- No standard/easy way to submit requests
- Requests for even more PII