Anthropic's AI R&D Doubles, But Evidence Hidden

🔥 What happened METR evaluated Anthropic's internal R&D and found the company is now improving model capabilities at 1.9x its historical rate since October 2025. Most of that acceleration comes from AI assisting Anthropic's own researchers, not from more compute. 💡 Why it matters Anthropic packed two years of historical progress into one – and METR sees no clear bottlenecks for the next 12 months. If AI keeps accelerating its own development, capabilities could spike suddenly before anyone can react. ⚡ Our take Anthropic redacted the most informative evidence, claiming it could help competitors. That's not transparency – that's a PR move. If you're going to drop numbers this alarming, show the receipts.