METR: Anthropic's internal models show 1.9× faster AI R&D gains
anderssandberg · x · 2026-11-08
Sharing METR's independent assessment of AI R&D acceleration at Anthropic, Anders Sandberg says he's "not fully convinced the research rate doubled, but not impossible either." METR finds benchmark scores (AECI aggregate) growing at 1.9× the pre-October 2025 rate (95% CI 1.5×–2.8×) once Anthropic's recent internal models are included — the trend jumped from +15/yr to +29/yr after Oct 2025. Anthropic redacted most of the evidence METR found most informative, citing confidentiality that could accelerate rival developers, leaving METR and Anthropic researchers themselves with significant uncertainty about the holistic picture.
More from Safety
- Indiana city unanimously bans recursively self-improving AI within city limits — austinc3301 · 2026-11-08
- Nous Research's Propaganda_Eval Dataset Trends on Hugging Face — NousResearch · 2026-11-08
- Leeds AI datacentre row: water use equals just 5 car washes, 1,000+ jobs at stake — danbri · 2026-11-08
- Nina Schick: Superintelligence risk hinges on humans controlling humans and sovereignty — NinaDSchick · 2026-11-08
- Asahi Linux bans generative AI use in project development — Pyros-SD-Models · 2026-11-08
- Australia moves to restrict screens and AI in schools, drawing criticism it will leave kids behind — CtrlAltDwayne · 2026-11-08