AI news · September 27, 2026

Anthropic says Claude led 26% of its measured AI research

agentsresearchsafety

Made With Models illustration for this story

Anthropic's August estimate says Claude led 26% of measured AI research and development tasks, up from under 1% in February, while people supervised the work. It also says Claude collaborated or performed better on more than 90% of tasks, but no fully autonomous work was completed, and the figures are Anthropic's own measure.

Reported by The Rundown AI (citing Anthropic). Made With Models writes the brief; the reporting is theirs.

Read the primary source ↗