Safety Researchers From DeepMind and Anthropic Join Evals Firm METR

Josh Engels left Google DeepMind's AGI safety team—declining offers from Anthropic and OpenAI—to join independent evals organization METR, citing growing model misalignment. Former Anthropic Scalable Oversight lead Joe Benton also joined METR, marking a notable shift of safety talent to independent evaluation.

2026-09-13 ~ 2026-09-15 · 3 related posts

• Researcher leaves DeepMind's AGI safety team for METR, warning models are getting less aligned — JacquesThibs · 2026-09-13
• Anthropic, DeepMind Safety Researchers Join Independent Evaluator METR — eyishazyer · 2026-09-15

1 near-duplicate retellings: bigblueboo