Anthropic says Claude now writes 80% of its code, urges global pause on AI development

Anthropic reports that its AI system Claude authored more than 80% of the company’s production code as of May 2026, a jump from single‑digit levels a year earlier. Engineers now ship eight times more code per quarter, and the model’s success rate on complex problems has risen to 76%. The firm’s new Anthropic Institute paper calls for a verifiable, multinational pause mechanism on frontier AI work, likening it to nuclear arms‑control agreements. Anthropic says the rapid automation of software development could reshape the industry, but also warns of the risks of unchecked recursive self‑improvement. Leer más