Pretraining researcher quits Anthropic: OpenAI and Anthropic are "gambling with our lives"
EthanJPerez · x · 2026-09-10
A researcher who spent three years doing pretraining research at both OpenAI and Anthropic has resigned, saying neither company is acting responsibly and both are "racing straight to self-improving superintelligence and gambling with our lives." @theo pushes back on accusations that Jacob is faking it for clout: he is highly regarded in the research community, backed by Anthropic coworkers, and has far more to lose than to gain. The resignation, paired with his "burn my equity for a 1% better survival odds" remark, marks a visible safety-vs-speed split inside frontier labs.
Related event: Anthropic safety researcher quits, warning of existential AI risk(119 posts)→
More from AGI Musings
- The short case for AI killing everyone: control, indifference, and the cost of keeping humans alive — DavidSKrueger · 2026-09-10
- Ex-Meta researcher cites Vernor Vinge's rogue AI escape as air-gap argument — seanwbren · 2026-09-10
- Satellite cobalt bombs as an AGI deadman switch? Twitter debates the deterrence idea — tszzl · 2026-09-10
- OpenAI employee on safety: with 1 billion weekly users, the responsibility weighs on everyone — Scobleizer · 2026-09-10
- Ex-Meta AI researcher warns OpenAI's unaligned agent swarm could cripple an entire nation — Polymarket · 2026-09-10
- Ex-Meta AI researcher warns OpenAI's unaligned agent swarm could 'cripple a nation' — Polymarket · 2026-09-10