Anthropic capabilities employee: AI poses moderate extinction risk
nabla_theta · x · 2026-09-11
An Anthropic employee stated publicly, in a personal capacity, that he believes there is a moderate chance of human extinction from AI. He works on a capabilities team because he considers Anthropic the most responsible actor in the space, and his primary motivation is reducing extinction risk.
A reply suggested he switch to safety work, highlighting the ongoing debate over mitigating risk from within capabilities teams versus dedicated safety roles.
More from AGI Musings
- Professor: STEM education now exists to keep students ignorant of AI they'll be judged by — RexDouglass · 2026-09-11
- If AGI Emerges at Competing Companies, Should Aligned AI Shut Down Its Rivals? — Diligent-Buy-5428 · 2026-09-11
- Big AI labs' researchers fear extinction but fear losing the race more, researcher says — Yuchenj_UW · 2026-09-11
- Philosopher Tyler John urges academics to tackle real AI problems, not niche topics — RishiBommasani · 2026-09-11
- Viewing LLMs as threat and opportunity for entrenched software incumbents — NickPassig · 2026-09-11
- iamtrask: concentrating AI power in 3 labs isn't an x-risk mitigation — we need federation — iamtrask · 2026-09-11