The US and China are set to compare notes on AI safety this week, even as one keeps choking the other's chip supply.
Treasury Secretary Scott Bessent says the two countries have discussed a new notification channel, a hotline for flagging when an AI system in either country starts acting unpredictably or poses a safety risk. Bessent framed it as a step toward sharing a "vision of common goals" on AI. The idea surfaces ahead of a Trump-Xi meeting scheduled for Thursday and Friday, September 24 and 25, 2026. It would require China's active participation to function, and Beijing has not publicly confirmed it is on board.
That silence matters. Trump has spent the year expanding export controls aimed at slowing China's AI progress, while Congress weighs bills to cut off even more chips and hardware. Add this month's Justice Department accusation that six major Chinese AI firms stole research from US frontier labs, a claim Beijing disputes, and asking China to jointly police AI risk while limiting its access to the technology that creates that risk looks like a tough sell.
A safety hotline is a reasonable idea on its own merits. Whether it survives contact with an escalating chip war is the real story here, not the announcement.