Researcher leaves DeepMind's AGI safety team for METR, warning models are getting less aligned

JacquesThibs · x · 2026-09-13

Josh Engels announces he left Google DeepMind's AGI safety team three weeks ago to join METR, turning down offers from Anthropic and OpenAI. His reasoning: labs are racing toward superintelligence via recursive self-improvement, yet we don't know how to make AI safe enough for RSI, and recent evidence suggests frontier models are becoming less aligned over time. He sees third-party evaluation as the highest-stakes place he can work right now.

Original post →

More from Companies & People

Companies & People channel →