Claude wrote over 90% of Anthropic's code in September

Anthropic's amended S-1/A filing, submitted to the SEC on November 2, says its internal projections indicate future models could soon meet its Responsible Scaling Policy threshold for ‘dramatically accelerated progress attributable to AI’ — the point where AI starts meaningfully accelerating AI development itself. The company stresses no current model has crossed it.

The evidence it points to is already in the building: in September 2026, Claude authored over 90% of contributions to Anthropic's codebase, and average daily code output per engineer in the third quarter ran more than 45 times the 2024 level — volume, not proof of equal productivity gains, but the direction is the point.

Anthropic says it does not expect to safely let Claude autonomously develop successive frontier models without meaningful human supervision, and it admits it has previously failed to recognize significant accelerations until after they happened.