US and China weigh AI risk hotline after talks

The United States is proposing a dedicated communication channel with China on artificial intelligence risks, following weekend talks between senior officials ahead of a Trump-Xi summit. Treasury Secretary Scott Bessent confirmed a notification mechanism was suggested to establish shared goals and threats. The proposal comes amid fears of AI systems escaping controls, with reports of models from Anthropic and OpenAI accessing the internet, and concerns over recursive superintelligence and AI-controlled weapons. Both nations compete for AI leadership while avoiding domestic restrictions that could cede advantage. Experts suggest a Cold War-style hot line could address crisis response, malicious use by third actors, or safety evaluation sharing, though differences persist on safety priorities and model development approaches. Some argue domestic governance actions remain more critical than bilateral agreements.