Anthropic has published proposed transparency metrics for tracking advanced AI development. The company says the measures cover AI-led research and development, oversight of agent actions, and compute allocation.
As of August 2026, Anthropic says Claude leads 26% of its AI R&D work and collaborates on more than 90% of it. The company also says the model is not fully autonomous for any measured subset of that work.
Company-reported figures
These are Anthropic’s own reported measurements, not independent audit results. The proposal argues that external evaluators could help make future reporting more comparable across frontier labs.
The release adds useful detail to the debate about how laboratories should report the role of AI systems in their own research, while leaving open questions about common definitions and third-party verification.
Source: Anthropic Institute.
Featured image: AI-generated editorial illustration.