TechWhy a new report on Anthropic automating AI development is alarming researchers

An 11-week METR investigation found Anthropic’s models are improving roughly twice as quickly as a year ago, with most of the acceleration attributed to AI rather than increased training compute. AI mainly assists human researchers for now, though METR sees signs autonomous contributions may grow soon. Anthropic’s extensive redactions also raise questions about public insight into frontier AI’s pace and risks.

AIAnthropicMETRAI safety

TThe Verge★★★☆☆2026-11-07 12:30Original