China — AI regulation
A third governance model the wiki’s horizontal-vs-sectoral fork lacked: binding, content-focused, state-driven regulation. China didn’t wait for a single horizontal act — it shipped early, targeted, enforceable rules ahead of the eu-ai-act, organized around state control of content and data rather than fundamental-rights risk tiers.
The instruments
- Algorithm Recommendation Provisions (March 2022) — recommendation algorithms must register with the Cyberspace Administration of China (CAC); by mid-2024 1,400+ algorithms were filed under this regime. (Compulsory algorithm registration has no EU/US analog.)
- Interim Measures for Generative AI Services (in force 15 Aug 2023) — “one of the first comprehensive national frameworks for generative AI”: security reviews, content control, plus data-protection/transparency/accountability duties on providers.
- AI-Generated Content Labeling Measures (2025) — mandatory on-screen disclosure + embedded provenance metadata + watermarking for AI text/image/audio/video. (A binding version of the voluntary SynthID-style provenance the sibling speech-audio-wiki tracks.)
Where it sits on the wiki’s axes
- Binding + early, but content-first. Like the eu-ai-act it is binding and broadly scoped, so it isn’t Israel’s sectoral soft law — but its organizing principle is state/content control + data localization, not the EU’s rights-and-safety risk tiers. So the wiki’s fork widens from two poles to a triangle: EU (rights-driven, horizontal-binding) · China (state-driven, content-binding) · Israel (innovation-driven, sectoral-soft) — with the US as a fourth, market-driven pole.
- Provenance as compulsion. China’s labeling mandate is the hard-law end of the AI-provenance spectrum, where most regimes (and vendors) sit at voluntary watermarking.
Related
eu-ai-act · us-ai-policy · responsible-innovation · risk-based-regulation · ai-governance · oecd-ai-principles