Anthropic says its model Claude is helping to build the next version of itself

Anthropic's AI model Claude is playing a key role in developing the next iteration of the company's own technology, the artificial intelligence lab announced on Thursday. Claude is responsible for 26% of Anthropic's model research and development, allowing the company to complete most tasks from a high-level prompt while still being under human supervision. The model is not yet fully autonomous, but it handles about 90% of the company's research and development in close human direction.

This announcement comes amid calls for a slowdown in AI development by leading figures in the field, primarily Anthropic CEO Dario Amodei. The company is advocating for better measurement and public reporting of AI development to minimize the gap between leading labs' knowledge and public understanding. Anthropic argues that sharing metrics on model development could provide a clearer understanding of how close AI systems are to achieving recursive self-improvement, or the ability to autonomously build their successors.

Anthropic has implemented agent oversight measures, with around 30,000 agents involved in research and engineering work as of August. External third-party evaluators will be embedded within the company to monitor safety efforts. However, the pace of Claude's contribution to research and development has raised questions about how close the company believes it is to achieving recursive self-improvement.