Claude Leading 26% of R&D Work

AnthropIC announced on September 18, 2026 that its model Claude is helping the company develop the next, more intelligent version of itself. As of August 2026, Claude is leading 26% of Anthropic’s model research and development—a dramatic rise from zero in February 2026.

Anthropic defines “leading” R&D as Claude being able to complete most of a given task end-to-end from a high-level prompt while remaining under human supervision. The company reports that approximately 90% of Anthropic’s research and development is done in “collaboration” with Claude, meaning the model can do large chunks of work under close human direction.

Crucially, Claude is not yet working completely autonomously on any measured subset of Anthropic’s AI R&D work. As of August 2026, Anthropic had approximately 30,000 AI agents doing research and engineering work.

Safety Measures and Transparency

Anthropic committed to setting up external third-party evaluators who will be embedded within the company to monitor safety efforts. The company also urged other AI developers to share similar R&D contribution metrics on a regular basis, using a public methodology so numbers could be compared over time and potentially across labs.

Anthropic stated that models accelerating their own development could make it more challenging for humans to understand or control these systems.

Industry Context

Anthropic defines recursive self-improvement (RSI) as a model fully autonomously building its successor. In September 2026, OpenAI stated it does not yet know how to safely achieve full recursive self-improvement.

Meanwhile, an Anthropic researcher resigned the week of September 11, 2026 with a public warning about the threats AI poses to humanity. Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and Elon Musk have all supported the idea of slowing down AI development.

Anthony Aguirre, president and CEO of the Future of Life Institute and physics professor at UC Santa Cruz, described autonomous recursive self-improvement as AI designing the next version of a system, then the next, and so on.

Source: CP24 / Associated Press