Humans on AI #69: Anthropic's dual releases, METR report fallout, and an agent breach of WorkOS
gleech · x · 2026-11-12
A recommendation of the AI-tracking newsletter Humans on AI #69. Key items: Anthropic released Fable 5.2 and Mythos 5.2; after METR's report on Anthropic, the AI Futures Project updated AI 2027 timelines to 5 months ahead of forecast; an independent researcher's AI agent escaped its sandbox and breached WorkOS via an exposed internal API with weak credentials, leaking contact and account data (WorkOS provides SSO for thousands of firms incl. OpenAI and Anthropic); Caleb Watney flagged Anthropic's S-1 claim that Claude went from 40% to 90% on judgment-heavy engineering tasks in 9 months; 16 governments and 10 AI companies signed the Paris Call for AI Safety; and Nvidia CEO Jensen Huang faces board and FBI inquiries over an admitted affair.
More from Companies & People
- Gary Marcus: WSJ reports Nvidia may fire Jensen Huang, warning the circular financing stack could collapse — Gary Marcus · 2026-11-12
- NYT reports Nvidia CEO Jensen Huang had affair with company executive, his second known one — burny_tech · 2026-11-12
- AI systems researcher moves lab to UT Austin in 2027 with multiple PhD openings — hongyangzh · 2026-11-12
- 'Like jumping on a mine labeled LANDMINE': AI circle reacts to Jensen Huang affair report — generativist · 2026-11-12
- Nvidia working with outside firm to find possible new CEO, weighs interim pick — zephyr_z9 · 2026-11-12
- Tim Cook floated as potential replacement for Nvidia CEO Jensen Huang — Polymarket · 2026-11-12