AI Security
LLM agents, cyber reasoning systems, ML-assisted vulnerability analysis, fuzzing, exploit generation, program analysis, red teaming, and scalable defenses.
Publications
All publicationsCOIN: A Live, Large-scale Benchmark for Semantics-to-Input Code Reasoning
CtxFuzz: Context-Aware Directed Fuzzing via Runtime Fuzzer-Agent Cooperation
SoK: DARPA's AI Cyber Challenge (AIxCC): Competition Design, Architectures, and Lessons Learned
OSS-CRS: Liberating AIxCC Cyber Reasoning Systems for Real-World Open-Source Security
ATLANTIS: AI-driven Threat Localization, Analysis, and Triage Intelligence System
Revisiting Binary Code Similarity Analysis using Interpretable Feature Engineering and Lessons Learned
Improving Large-Scale Vulnerability Analysis of IoT Devices with Heuristics and Binary Code Similarity
Talks
All talksAI Agents for Security: Lessons from AIxCC and AI Security Research
Inside the AI Cyber Challenge
AI Security 101: Red Team's Perspective on Navigating New Threats in AI Frontier
Red Team's Perspective on Navigating New Threats in AI Frontier
AI Security 101: Red Team's Perspective on Navigating New Threats in AI Frontier
AI Security Primer: Red Team Perspectives on Navigating New Threats and Safeguarding AI Frontier
Security for Generative AI, Generative AI for Security (Panel)
AI Security Primer: Red Team Perspectives on Navigating New Threats and Safeguarding AI Frontier
Blog
All blog postsHow Can You Trust a Model Someone Else Trained?
Running an open model on your own server doesn't tell you what's inside the weights. Finding and fixing model backdoors is following the path software security already took, but models are harder both to read and to fix.
AI Reverse Engineering Can Copy Software, but Vulnerabilities Need an Exact Look
There's a lot of talk these days about copying commercial software with AI, but what AI reads out of a binary isn't the original. It's code with the blanks filled in plausibly. That may be enough for copying, but when you're looking for vulnerabilities, those blanks become the problem.
Vulnerabilities Are Pouring In. Will Patches Reach Users in Time?
AI is producing vulnerabilities and exploits faster than ever, but patches aren't reaching users any faster to match. I took the patch gap and blast radius questions I raised in March and checked them against seven months of new numbers.
Getting Vulnerabilities Found by AI Scanners to Developers
AI has made finding bugs easy. Filtering them, fixing them, and handing them to maintainers still takes a lot of work.
When I Hand Work to an Agent, Who Makes the Judgment Calls?
However carefully you write down what you want, it's hard for the person you hand the work to to follow your thinking 100%. Agents are no different. Here's what needs to be aligned up front, who should review the code an agent writes, how much to automate, and who owns the judgment that's still left after you automate.
The Accelerating Gap Between Attack and Defense in the Age of AI
The time from vulnerability disclosure to exploit has shrunk from 2.3 years to 1.6 days. AI is widening the gap between attack and defense.
What If We Let AI Agents Pay?
Google's AP2, Cloudflare's NET Dollar, ChatGPT Instant Checkout: a roundup of the agent payment announcements that poured out within two weeks in September, with my notes on why agents end up spending money and what worries me about it.
Team Atlanta Wins DARPA AI Cyber Challenge (AIxCC)
Team Atlanta wins 1st place in DARPA's AI Cyber Challenge (AIxCC), demonstrating AI-driven vulnerability detection and patching.
Random Thoughts on GenAI
If an AI learns from the same data as me, how do I prove which one is the real 'me'? Random thoughts on fake information and phishing getting more convincing with generative AI, why AI red teams need to look at safety too, and a study that jailbroke LLMs with low-resource languages.