China Issues 3rd AI Safety Framework as It Rejects US Calls to Slow Development
Updated
Updated · The Guardian · Sep 20
China Issues 3rd AI Safety Framework as It Rejects US Calls to Slow Development
3 articles · Updated · The Guardian · Sep 20
Summary
China this week published the third version of its AI safety governance framework, adding risks from AI agents, frontier-model cybersecurity threats and recursive self-improvement.
Beijing is resisting US-backed slowdown arguments because it treats AI as a core sovereignty capability and sees pacing proposals as an attempt to preserve Washington’s lead.
Analysts said the new framework still shows overlap with US safety concerns, even as China’s foreign ministry attacked “fearmongering” and Donald Trump dismissed extinction risks as a hoax.
China’s approach pairs fast-moving regulation with aggressive deployment: it has had an algorithm registry since 2022, issued rules this year on companion AIs, and is pushing adoption across factories and consumer services.
150 million users on digital health app A-fu and support for firms from local governments underscore a broader Chinese view that AI should be mastered and managed, not slowed.