Anthropic says Claude now writes over 90% of its code and wants the world to have an AI pause button

Editor's take

Anthropic's internal metrics reveal Claude is now responsible for over 80% of the company's production code, enabling engineers to ship eight times more code daily compared to earlier this year. This significant acceleration in self-development highlights the increasing maturity of LLMs not just as tools for external users, but as integral components in the very creation of AI systems.

This development matters because it signals a potential inflection point in AI research and development velocity. If companies like Anthropic can achieve such rapid internal progress, it suggests a future where AI development cycles shorten dramatically, impacting competitive landscapes and the pace of innovation across the industry. The implications extend to the workforce, as AI assistants become more deeply embedded in the coding process.

The next crucial observation will be whether this internal coding efficiency translates to tangible improvements in Claude's capabilities or the development of entirely new AI architectures, rather than just increased output of existing functionalities. Furthermore, Anthropic's call for an "AI pause button," juxtaposed with this internal acceleration, raises questions about how the industry will manage the ethical and societal implications of such rapid, self-driven progress.