As AI leaders warn of catastrophe, US and China shun slowdown calls — SkimNews
Get the Geopolitics newsletter
Daily geopolitics — wars, elections, sanctions, the diplomatic moves that move markets. Free.
- Trump dismissed AI fears as a "conspiracy" and told the UN General Assembly Washington would oppose any "globalist scheme" to control AI, casting US dominance in the technology as crucial to maintaining global power.
- Xi Jinping and Chinese officials rebuffed calls for a slowdown, dismissing Silicon Valley's existential risk warnings as "fearmongering," even as China released the third iteration of its AI security framework this month.
- Chinese Minister of State Security Chen Yixin warned in a rare essay this month of "hostile forces" using AI to "fabricate political rumours" and threaten "political security, institutional security and ideological security."
- Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman called for a slowdown and greater regulation, while Nvidia CEO Jensen Huang and Meta CEO Mark Zuckerberg played down the need for external oversight.
- The only proposals publicly floated by either side are a US-China dialogue and a notification mechanism for national security threats—no agreement on capabilities or uses that should be off-limits has been put forward.
- Recursive self-improvement—when AI models learn to improve themselves—was identified by Otto Barten of the Existential Risk Observatory as the most critical tipping point requiring a US-China "red line" agreement.
- Analysts identified a fundamental mismatch in risk perception: US experts focus on highly capable AI escaping human control, while Chinese officials focus on the US gaining a "decisive AI advantage" and using it against China.
Why it matters: The US and China host nearly all frontier labs and most downloaded models, so bilateral guardrails are "close to a precondition" for safe AI—without them, any safety regime covers only "half the problem." Neither superpower has proposed red lines on dangerous AI capabilities, and the only concrete measures floated are a dialogue and threat-notification mechanism, leaving risks like recursive self-improvement entirely unregulated.
Ask SkimNews
