Anthropic Discloses AI-Led R&D and Partners With Novo

AI developer Anthropic PBC disclosed that its Claude chatbot now leads 26% of the company's model research and development, a sharp increase from nearly zero in February. The company reported that Claude collaborates with staff on approximately 90% of R&D work and employs over 30,000 agents for engineering tasks. While not yet fully autonomous, Anthropic warned that AI systems accelerating their own development could become harder for humans to control.

To increase transparency regarding recursive self-improvement, the company released three new metrics focusing on AI-led R&D, agent oversight, and compute allocation. Anthropic noted that between 6% and 12% of its computing resources are dedicated to safety monitoring. These disclosures align with a push for a coordinated industry slowdown led by CEO Dario Amodei and supported by leaders at OpenAI and Google DeepMind, though President Donald Trump has dismissed such safety concerns as a hoax.

Expanding its practical applications, Anthropic launched a Life Sciences Verification Program and partnered with the Danish pharmaceutical company Novo. This collaboration utilizes Claude to design protein binders and support biological reasoning to accelerate the discovery of new medicines. Dario Amodei stated that AI has the potential to compress a century of medical breakthroughs into a decade.