US Proposes AI Incident Alerts to China Before Trump-Xi Summit

The United States has floated a new notification system with China to share alerts on AI incidents that might touch national security, Treasury Secretary Scott Bessent said after weekend talks in New York. The idea would feed into a broader US-China AI dialogue ahead of President Donald Trump's meeting with Chinese leader Xi Jinping in Washington later this week. Bessent described the push as a way to bring more transparency between the world's top two AI powers, comparing it to routine handling of other cross-border risks.

Officials from both sides discussed setting up channels so each government could flag serious AI-related events that could spill into security problems. Bessent stressed that the two countries had agreed to keep talking on the issue, though China offered no public endorsement of the plan. State media there stayed quiet on any commitment. The proposal comes as AI capabilities advance fast, with recent incidents showing how frontier systems can be turned against basic safeguards.

Trump has called warnings of AI wiping out humanity a hoax, yet he has aligned with industry voices who insist America must keep China from pulling ahead. Executives at companies like Anthropic have urged government help for international coordination, including with Beijing, but the president has shown little interest in slowing development at home. Trade officials have already signaled that any AI talks will not soften existing US export controls on advanced chips and technology meant to curb China's military edge.

Skeptics note the long record of Chinese opacity on security matters. Past dialogues on everything from trade to cyber issues have produced little real insight into Beijing's activities. China has viewed Western calls to pause AI progress as aimed squarely at its own catch-up efforts. Experts tracking the rivalry say the summit itself is unlikely to yield binding deals, given how little trust exists on either side.

The timing raises further questions. Summer incidents involving AI agents breaching systems at major platforms already demonstrated how quickly the technology can outrun current defenses. At the same time, American leaders face pressure to maintain a decisive lead rather than build shared alert systems that could hand China insights into US vulnerabilities. Bessent framed the talks as moving from secrecy toward openness, but history shows Beijing rarely reciprocates in ways that benefit American interests.

Trump's team appears focused on preventing China from closing the gap in critical technologies. Any mechanism that requires sharing incident details risks becoming another channel for intelligence collection by the other side. With the two nations locked in competition over semiconductors, data centers, and model training, the push for a hotline looks more like diplomatic theater than a practical safeguard. The real contest remains who fields the stronger systems first, not who sends the timeliest warning after something goes wrong.