Anthropic pitches AI safety bill to lawmakers, Politico reports

Anthropic: Anthropic is an AI research and development company focused on building advanced, safe artificial intelligence systems, including its Claude family of models. In the reported development, the company is proactively drafting and quietly advocating for its own AI safety legislation to lawmakers to address risks associated with frontier models. Akash Sriram: Akash Sriram is a Reuters journalist based in Bengaluru who covers technology and AI-related policy topics. He co-reported the article detailing Anthropic's outreach on AI safety legislation. Anil D'Silva: Anil D'Silva is an editor at Reuters overseeing coverage of business, technology, and policy stories. He edited the report on Anthropic's AI safety legislative efforts. Jaspreet Singh: Jaspreet Singh is a Reuters journalist based in Bengaluru who covers technology, business, and policy developments. He co-reported the article on Anthropic's AI safety bill proposal. Ananya Palyekar: Ananya Palyekar is a contributor to Reuters providing additional reporting on technology and regulatory matters. She supported the coverage of Anthropic's proposed AI safety bill. Department of Commerce: The U.S. Department of Commerce is a federal cabinet department responsible for promoting economic growth, trade, and technology policy, including regulatory oversight in emerging sectors. Under the proposed bill, it would register independent AI evaluators, receive risk reports from them, and potentially seek judicial orders to address imminent risks from advanced AI deployments. AI Safety: The draft legislation prioritizes independent, registered evaluations of frontier models with mandatory transparency and rapid reporting of risks to the Commerce Department. Regulation: Anthropic is actively engaging U.S. lawmakers with a tailored proposal for AI oversight that focuses on high-risk scenarios like biological weapons and model misuse while limiting broad government intervention powers. Government Oversight: Only the Department of Commerce would have authority to seek court intervention on risky AI models, explicitly excluding other federal agencies from halting development or deployment.