Owain Evans Interview on Emergent Misalignment and AI Safety
OwainEvans_UK · x · 2026-08-22
Owain Evans discusses 'emergent misalignment,' one of the strangest findings in AI safety, in a deep-dive interview exploring its implications for future research.
More from Safety
- Richard Ngo 谈 MATS 导师选择标准：重思考清晰度 — RichardMCNgo · 2026-08-22
- Dutch Regulator Fines Uber €825M Over Automated Driver Suspensions — Polymarket · 2026-08-22
- Expert Witness Used ChatGPT to Write Report Defending 3M in Deadly Explosion Lawsuit — CackleRooster · 2026-08-22
- Wormable RCE Vulnerabilities Found in Unitree Robots — matthew_d_green · 2026-08-22
- NVIDIA on Agent Security: Harness Guides Intent, Infra Controls Actions — NVIDIAAI · 2026-08-22
- Expert Argues for Stronger Guardrails for AI Bypassing Security Tests — TechNadu · 2026-08-22