Open Problems in Securing Systems Against AI Agent Swarms | John Kitaoka

​HackTalk: John Kitaoka

​John Kitaoka is a research fellow at MATS Research working on dangerous capability evaluations and scalable oversight for multi-agent systems with UK AISI. Prior to MATS, John worked with scientists at Harvard and Lawrence Livermore National Labs on automated red-teaming and verification techniques for dangerous knowledge elicitation by frontier models. He spent four years at Amazon Web Services as an AI Architect, and led AI product development for government technology sectors. His work spans scalable LLM evaluation for enterprises, multi-agent systems for critical sectors, and advising AI adoption across the public sector, and John has a background in mathematics and computer science from the University of Wisconsin-Madison.

​Talk: “Open Problems in Securing Systems Against AI Agent Swarms”

​Format: 15-30 min talk + 10-15 min Q&A. Online (Zoom); the Zoom link is shown once you RSVP.

​Part of the Swarm Alignment Hackathon, November 6-8, online and at AE Studio in Los Angeles and seven other local hubs around the world. Co-organized by Apart Research and AI Alignment Foundation in partnership with AE Studio, sponsored by Adaption.