What happened
Before Trump and Xi meet Thursday and Friday, Treasury Secretary Scott Bessent said both sides discussed an AI safety notification mechanism, an open line for alerting each other when AI systems pose threats or behave unpredictably.
Why it matters
The proposed channel could let the two countries share early warnings on dangerous AI behavior, but whether it becomes a reliable global governance framework depends on how much the rivals actually cooperate.
What to watch
The plan includes no officials with technical expertise to assess urgency, and one source mocked it as "BS" — a sign the talks may not deliver meaningful safety cooperation.
WHO IT HITSThis lands on US and Chinese government officials negotiating AI policy, as well as companies and researchers whose cross-border AI safety work hinges on whether the two countries build a trusted alert channel.
Summaries like this, in your inbox every morning.
The AI safety notification mechanism comes amid rising US-China tensions. Trump has steadily increased export controls aimed at blocking China from advanced AI technology, and Congress is considering bills that would cut off more chips and hardware. Two weeks ago, Trump's Justice Department accused six of China's leading AI makers of stealing from US frontier labs, which China strongly disputed.
Despite Trump's Truth Social posts insisting America must dominate China on AI, China showed up to the talks. Both countries already have another meeting scheduled in November, which Alvin Wang Graylin, a digital fellow at Stanford Digital Economy Lab, called encouraging. He suggested the US and China share many AI risks, from hackers or terrorists wielding AI to threats against infrastructure.
However, the mechanism's effectiveness may hinge on whether the US continues tying AI safety talks to keeping America ahead. Sam Bresnick of the Center for Security and Emerging Technology warned that if China feels "AI safety" is a code word for suppressing Chinese AI, the alert system could become useless in the long term. Graylin argued the US is caught up in a "deeply flawed" AI race narrative that obstructs cooperation on real imminent risks.
For example, today's edition would include:
AI-summarized, only the topics you pick: one digest a day via Email, LINE, or Slack.
Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →
Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.
