Anthropic Discloses Claude Leads 26% of AI R&D Work

Anthropic announced on Thursday that its AI model Claude now leads 26% of the company's artificial intelligence research and development work, a rapid increase from less than 1% in February. The disclosure provides a detailed look into how AI systems are increasingly involved in building next-generation models, while still operating under human supervision. According to Anthropic, Claude's role is defined as leading work where the model can complete most tasks end-to-end from high-level prompts, with humans primarily supervising. As of August 2026, over 90% of Anthropic's AI R&D involves Claude at a collaborative level or higher. The company emphasized that Claude has not reached full autonomy in any measured area. Anthropic also revealed that approximately 30,000 AI agents were performing research and engineering work on its internal platform in August. These agents' actions are monitored, with an automated system blocking about 1 in 47,000 decisions. A second monitor flags around 100,000 transcripts weekly, with roughly 50 reaching human reviewers.