Anthropic Researcher Quits Over Extinction Risk; Newsletter Writer Pushes Back
alex_verem · x · 2026-09-19
Jacob Coxon, a pretraining researcher who left OpenAI and then Anthropic, publicly resigned claiming both labs are racing to self-improving superintelligence, saying builders themselves believe AI could be lethal by decade's end — the post drew 170M+ views, and a still-employed colleague put the odds above 10%. Newsletter author Alex Prompter recaps the week and argues the doomer logic is really compound interest: the same stacking curve can be used productively at work. Includes a copy-paste 'Compounding Error Check' prompt for auditing error accumulation in AI workflows, plus four links on both sides of the debate.
More from AGI Musings
- Minerva vs Casado: air-gapped model coordination still infeasible at 0.002 bps covert bandwidth — BlancheMinerva · 2026-09-19
- Economists Ben Moll and Alex Imas: don't expect double-digit GDP growth from AI within 10-15 years — binarybits · 2026-09-19
- AI Growth Bet: Per-Capita GDP Growing 15% Above Peak Has Just 0.9% Historical Odds — RishiBommasani · 2026-09-19
- AI Makes Old Expertise Cheap—and Creates More Expert Work Finishing What It Starts — every · 2026-09-19
- "Think AI Safety Is Bad? Wait Until the Public Writes AI Policy" — NathanpmYoung · 2026-09-19
- Many New Projects Are Being Built on Chinese Open-Weight Models — signulll · 2026-09-19