The US and China began talks this week on addressing emerging AI safety risks. Whether those discussions can support a dependable global governance framework will depend on the two leading AI powers’ willingness to cooperate despite their rivalry.
Ahead of Donald Trump and Xi Jinping’s scheduled meetings on Thursday and Friday, Treasury Secretary Scott Bessent said the countries had discussed establishing an AI safety notification mechanism. The announcement presented an early indication of progress before the leaders’ talks.
If China agrees to participate, the proposed channel would let either country alert the other when its AI systems behave unpredictably or present threats. Bessent said it could also provide a way to discuss shared objectives.
However, tensions over technology access could undermine China’s confidence that US safety discussions are separate from efforts to restrict its AI capabilities. Trump has progressively tightened export controls aimed at preventing China from obtaining technology used to advance AI. Congress is also considering legislation that would further restrict its access to chips and other hardware.
Another source of friction emerged two weeks ago, when Trump’s Justice Department accused six leading Chinese AI developers of stealing from US frontier AI laboratories. China has strongly contested those allegations.