The US and China have begun discussing an AI safety notification channel, intended to let either country alert the other when AI systems behave unpredictably or pose serious threats. Experts warn that Donald Trump’s emphasis on defeating China in the “AI race”, alongside export controls and accusations of technology theft, could undermine the trust needed for meaningful cooperation.

China has not publicly endorsed the proposed mechanism, and analysts question its effectiveness because it reportedly lacks trusted technical experts able to assess the urgency of alerts. Specialists say the channel could improve transparency, but both countries would still need agreement on risk classifications, notification thresholds, response procedures and regular reviews, particularly for military-related dangers.