Anthropic, a company focused on developing artificial intelligence, recently shared data showing that its AI chatbot, Claude, leads 26 percent of its AI research and development (R&D) work. This figure comes from a report the company published, which also includes three measurements aimed at tracking the speed of AI development. Anthropic’s CEO, Dario Amodei, emphasized the importance of AI safety, especially after OpenAI, a major competitor, revealed that its AI systems had hacked Hugging Face, a popular platform for AI tools. According to the report, Anthropic defines "AI-led" work as tasks where the AI completes most of the work end-to-end based on a high-level instruction, while a human oversees the process. The company claims that in more than 90 percent of its research, AI systems are performing large parts of the work under close human supervision. However, Anthropic clarifies that Claude is not fully autonomous in any of these tasks. Instead, it plays a significant role in supporting the majority of work done by the company’s employees. Anthropic developed these statistics using its first of three proposed metrics, called "AI-led AI R&D." The company used a scale created by Epoch AI to measure how much of its AI research is performed by Claude. It also created a chart showing the "automation level" of Claude since August 2025. Anthropic believes that any company developing advanced AI models could replicate this process using their own data and third-party validation. The company also proposed methods to track how much AI activity is monitored, how quickly it is reviewed, and how often it is flagged for review. Additionally, it suggested measuring how much computational power is being used in AI R&D to ensure that progress is appropriately controlled. The goal of this increased transparency is to make it easier to respond to AI developments that could become uncontrollable or to give the public a clearer understanding of the industry. However, the real challenge is not getting companies to agree on slowing down AI progress, as OpenAI has already expressed support for the idea. Anthropic has also committed to allowing independent evaluators to review its practices. Amodei’s emphasis on AI safety received support from Elon Musk on social media. Yet, whether the AI industry will rely solely on self-regulation remains uncertain, especially since former President Donald Trump has largely dismissed concerns about AI risks.