Anthropic has published three measurements of how much of its own AI research and development is done by AI, reporting that Claude “leads” 26% of that work, the company said on 17 September 2026.
On Anthropic's scale, “AI leads” means a model can complete most of a task end to end from a high-level prompt while a person supervises. As of August 2026, the company said, Claude was not operating fully autonomously for any measured subset of its AI R&D work, and more than 90% of the work was at or above the level where AI collaborates.
Anthropic said about 30,000 agents were doing research and engineering work at any one time on its most-used internal platform. Across more than a billion agent decisions in August, its online monitor blocked 0.002%, or about 1 in 47,000.
The ratings were made by a judge model. Its exact agreement with human raters was 59%, against 35% between humans, and model and human ratings were within one level of each other 97% of the time. Anthropic said there is still room for disagreement on borderline cases.
The company argued that any frontier developer could publish these measures regularly using a public methodology.