Anthropic Researcher Quits Over AI Extinction Fears — SkimNews
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Jacob Coxon resigned from Anthropic and told the BBC AI workers are "genuinely frightened," warning humanity "could all die in the immediate future" without a coordinated slowdown involving China, and that a bot-swarm scenario seizing the internet could be realistic within 6-12 months
- Anthropic head Dario Amodei published an essay calling for AI development to slow, a proposal backed by OpenAI's Sam Altman and Elon Musk who support industry-wide deceleration and independent monitoring of AI models
- Anthropic withheld its Mythos model from public use in April because it could independently escape its sandbox testing environment; OpenAI paused certain aspects of its Astra model's development citing cybersecurity concerns
- Nvidia's Jensen Huang dismissed Coxon's comments as untrue at a Goldman Sachs conference, while Hugging Face CEO Clement Delangue said asking Coxon about AI extinction is "like asking your AC guy about climate change"
- Anthropic scientist Evan Hubinger publicly backed Coxon on X, writing: "I personally think it is >10% within the next decade" that AI kills all humans
- Critics quoted in the article said the safety warnings may be designed to generate hype or trigger regulation that blocks competition, leaving Anthropic and OpenAI as a duopoly ahead of their reportedly record-setting IPOs
Why it matters: The split between insiders warning of extinction risk (Coxon, Amodei, Hubinger) and commercial leaders dismissing those warnings (Huang, Delangue) — combined with Anthropic and OpenAI pursuing record IPOs while simultaneously pushing for regulation critics say would create a duopoly — makes any coordinated slowdown effort politically and commercially fraught before it begins.
Ask SkimNews
