Anthropic Claude Leads 26% of AI R&D
Analysis based on 51 articles · First reported Sep 17, 2026 · Last updated Sep 18, 2026
The disclosure reinforces the narrative that leading AI labs are rapidly automating their own model development, which could accelerate capability gains and intensify competition among frontier AI companies, supporting valuations across the AI and semiconductor supply chain. At the same time, the emphasis on recursive self-improvement and safety oversight keeps regulatory and safety concerns in focus, a potential source of volatility for AI-linked equities.
Anthropic announced in a Thursday blog post that its AI model Claude now leads 26% of the company's artificial intelligence research and development work, up from below 1% in March on a scale developed by Epoch AI. The company said Claude can complete most of a given task end-to-end from a high-level prompt while remaining under human supervision, and that AI collaborates with humans on more than 90% of its research work as of August. Claude is not operating fully autonomously in any measured subset of AI R&D work. Anthropic said about 30,000 AI agents were performing research and engineering tasks on its main internal platform at any given time in August, with every action screened before execution; of more than 1 billion decisions that month, about one in 47,000 was blocked, and roughly 50 high-priority cases are escalated for human review weekly. The company also disclosed that about 6% of its AI research computing power went to safety work in a sample week in July, rising to 12% for research carried out by AI itself. Anthropic plans to publish these measurements regularly and urged other frontier labs to share comparable metrics verified by third parties, framing the disclosure as a way to gauge how close the industry is to recursive self-improvement. The announcement came amid an ongoing AI safety debate: an Anthropic researcher resigned last week with a warning about the technology's risks, and CEO Dario Amodei has called for frontier labs to coordinate on slowing development, a position supported by Sam Altman and Elon Musk but opposed by others including President Donald Trump. OpenAI separately said it would regularly publish reports on unexpected or unauthorized AI behavior.
Set up alerts, explore entity relationships, search across thousands of events, and build custom intelligence feeds.
Open Dashboard