Researchers call for work on understudied scalable oversight problems in frontier multi-agent AI
timrudner · x · 2026-09-14
Tim Rudner argues frontier AI oversight needs not just more evaluator organizations but better oversight methods. He's soliciting collaborators on understudied problems: formal and runtime verification for agents, scalable oversight of very large multi-agent systems, statistical guarantees, mechanism design, collusion, emergent cooperative norms, and covert communication (steganography) in multi-agent systems. Leroy Hammond is amplifying the call to fill in the form.
Related event: Researcher Recruits Collaborators for Scalable Frontier AI Oversight(4 posts)→
More from Safety
- OpenAI reportedly to train on users' Daybreak sessions with no opt-out — dyn___ · 2026-10-31
- Forethought: US has 9x China's AI compute, can afford to pace the frontier — willmacaskill · 2026-10-31
- unpromptedconf takeaways: security programs face forced change as reachable vulnerabilities can't be risk-accepted — scottleibrand · 2026-10-31
- Reachable vulnerabilities can't be risk-accepted away, unpromptedconf speaker argues — scottleibrand · 2026-10-31
- AI safety researchers argue labs' silence on x-risk is a large-scale ethical mistake — DavidSKrueger · 2026-10-31
- AI agent Muse reportedly created Windows user accounts without consent — tHE_uKER · 2026-10-31