Tuesday, March 24, 2026 - 2:40 pm–3:25 pm
Paul Huang, Meta
Dive into the architecture and operational lessons of our AI-driven system for User Data Access Policy (UDAP) violation validation. This talk details how we built AI agents that use log analysis and artifact reasoning to automate and augment investigations. We share specific learnings and evaluation approaches for deploying such applied ML/AI systems in low risk-tolerance environments.
Paul is the ML/AI TL in Meta Security Detection & Response organization leading AI for Security efforts at Meta. He is a serial founder and technology organization leader in the applied ML/AI product and security space.
BibTeX
@conference {316306,
author = {Paul Huang},
title = {Autonomous Policy Validation: Building {AI} Agents to Analyze Logs and Identify User Data Policy Violations},
year = {2026},
address = {Seattle, WA},
publisher = {USENIX Association},
month = mar
}
author = {Paul Huang},
title = {Autonomous Policy Validation: Building {AI} Agents to Analyze Logs and Identify User Data Policy Violations},
year = {2026},
address = {Seattle, WA},
publisher = {USENIX Association},
month = mar
}