Anthropic released data showing that Claude, its AI chatbot, now leads 26% of the company's research and development work—a jump from essentially zero at the start of the year. This matters because it's one of the first concrete numbers we have on whether AI systems can actually speed up their own development. The company measured this by tracking which projects Claude takes the lead on versus which ones humans drive. Beyond just leading projects, Claude collaborates with Anthropic staff on roughly 90% of their work. That's a different metric than the 26% leadership figure—it suggests the tool has become woven into nearly all R&D activity, even when humans are still calling the shots. The distinction matters: collaboration is easier to achieve than having an AI system independently own a research direction. The timing is significant. Anthropic published this through its institute in a formal report, suggesting the company sees this as worth documenting and sharing publicly. Whether this trend continues, accelerates, or plateaus will shape how fast AI development actually moves. If AI systems can genuinely speed up their own improvement, we're looking at a different growth trajectory than if they're just useful tools humans happen to use more often.