HackTalk: Usman Anwar
HackTalk: Usman Anwar
Usman Anwar is a final-year PhD student in the Computational and Biological Learning Lab at the University of Cambridge, supervised by David Krueger and funded by the Open Phil AI Fellowship and the Vitalik Buterin Fellowship on AI Safety. His research focuses on AI safety and alignment, most recently chain-of-thought monitorability and detecting steganography in LLMs. He is the lead author of "Foundational Challenges in Assuring Alignment and Safety of Large Language Models" (TMLR 2024).
Talk: “Challenges in Detecting Steganography in LLM Transcripts”
In his words: “I will talk about why classical steganography theory is difficult to apply to LLMs and a potential workaround based on our work (arXiv 2602.23163).”
Format: 15-30 min talk + 10-15 min Q&A. Online (Zoom); the Zoom link is shown once you RSVP.
Part of the AI Collusion Research Sprint, October 23-25, online and at hubs in New York City and Hong Kong. Co-organized by Apart Research, Poseidon Research and AE Studio, sponsored by Modal, The AI Alignment Foundation and Nebius.