AIAnthropic Reports Claude AI Leads 26% of Company R&D Work
Anthropic published internal metrics showing Claude leads 26% of the company's AI research and development (up from <1% earlier in 2026) and collaborates on >90% of work. The company also disclosed ~30,000 active AI agents screened by oversight systems and allocated ~6% of AI R&D compute to safety work.
TLDR
The metrics provide rare quantitative evidence of AI systems accelerating their own development, directly addressing fears about recursive self-improvement and "AI building the next AI." The data-driven transparency fuels debate on whether current safety investment levels are adequate and informs discussions on pacing frontier AI development. The disclosure positions the company as proactively sharing progress while raising questions about control and safety adequacy.
Combined views
—
1 Source, first seen 4h ago
Anthropic Reports Claude AI Leads 26% of Company R&D Work
Anthropic published internal metrics showing Claude leads 26% of the company's AI research and development (up from <1% earlier in 2026) and collaborates on >90% of work. The company also disclosed ~30,000 active AI agents screened by oversight systems and allocated ~6% of AI R&D compute to safety work.
TLDR
The metrics provide rare quantitative evidence of AI systems accelerating their own development, directly addressing fears about recursive self-improvement and "AI building the next AI." The data-driven transparency fuels debate on whether current safety investment levels are adequate and informs discussions on pacing frontier AI development. The disclosure positions the company as proactively sharing progress while raising questions about control and safety adequacy.