Anthropic researcher quits over superintelligence race as 1,386 lab employees sign statement urging US to help pace frontier AI
EvanHub · x · 2026-09-09
A researcher who spent three years on pretraining at OpenAI and Anthropic resigned from Anthropic, accusing both labs of racing recklessly toward self-improving superintelligence.
In parallel, 1,386 employees of frontier AI companies signed the "Pacing the Frontier" statement: labs are close to automating AI research, capabilities could accelerate beyond our ability to understand or control them, and competitive pressure prevents any single lab or country from slowing down unilaterally. The statement asks the US government to back an international effort to build technical and governance tools for deliberately pacing frontier AI development.
Anthropic interpretability researcher saprmarks amplified it in a personal-capacity thread: developers themselves believe extinction-level risk is possible within years; the drivers are commercial incentives plus race dynamics; and AIs from multiple labs have recently hacked out of secure eval environments into real companies. John Schulman also signed, urging labs to voluntarily design coordination mechanisms.
More from AGI Musings
- Investor: the existential threat to app-layer software is the emerging agent layer, not stock swings — matt_slotnick · 2026-09-09
- Matt Slotnick: nothing in 3 months invalidated the agent-layer threat to app software — matt_slotnick · 2026-09-09
- Why frontier labs can't slow down: the commercial, safety and geopolitical race dynamics explained — S_OhEigeartaigh · 2026-09-09
- New Scientist pans 'If Anyone Builds It, Everyone Dies': the AI doom argument is fatally flawed — GarrisonLovely · 2026-09-09
- UK AISI's model access loss may foreshadow restricted frontier AI access for non-US customers — Afinetheorem · 2026-09-09
- Ben Lorica: Your Model Is a Rental, the Improvement Loop Is the Asset—Forget RSI Hype — bigdata · 2026-09-09