China's recommendation algorithm rules take effect

China’s Internet Information Service Algorithmic Recommendation Management Provisions took effect, becoming one of the first binding national regulations anywhere aimed specifically at recommendation algorithms. The rules were jointly issued on 31 December 2021 by the Cyberspace Administration of China alongside the Ministry of Industry and Information Technology, the Ministry of Public Security and the State Administration for Market Regulation, then took effect two months later.

The provisions required platform operators whose services carried “public opinion properties or social mobilization capacity” to file details of their recommendation algorithms with regulators within ten working days, including a self-assessment of the algorithm’s effects, through a dedicated registration system. Providers had to regularly examine and assess their algorithms’ outcomes and were barred from designing models that encouraged addictive or excessive use. Users gained a mandated right to opt out of algorithmic personalisation entirely, and to view, select or delete the personal tags platforms had assigned them for recommendation purposes.

The rules also targeted specific practices that had drawn public complaint in China: they banned “big data price discrimination,” where platforms charge different users different prices for the same service based on inferred willingness to pay, and set out protections for minors, elderly users and gig workers subject to algorithmic task allocation and scheduling — a direct response to controversy over delivery-platform algorithms setting increasingly tight delivery times. Violations carried fines of 10,000 to 100,000 yuan. Services that generated synthetic news content or deployed monopolistic algorithmic practices faced additional restrictions.

The provisions predated, by more than a year, the wave of Western AI legislation that followed ChatGPT’s release, and gave Chinese regulators an early operational template — mandatory filing, algorithmic transparency to the state, and user-facing opt-outs — that shaped China’s subsequent, more AI-specific rules on generative content and deep synthesis.