China regulates AI through a layered set of instruments: the Interim Measures for Generative AI Services, the algorithm recommendation rules, the deep synthesis provisions and, since 2025, the measures on labelling AI-generated content. Each carries a registration or filing duty, and each is enforced through platform inspection and the periodic publication of filed services.

This section sets out those duties as they apply in practice — what must be filed, which documents carry the weight, how long a filing takes, and what follows when a service changes. It is written for compliance and product teams that need a decision rather than a summary of the statute.

Key topics include algorithm filing, generative AI filing and registration, deep synthesis filing, large-model security assessment, content labelling, face recognition filing, on-device AI registration, and the enforcement campaigns that test them.

Writing the Security Assessment Report for Large-Model Filing: Eight Chapters and an Evidence Chain

The security assessment report for large-model filing is not a template to be filled in, but a case that must persuade the reviewer of three things: what the model can do, which risks those capabilities carry, and what evidence shows each risk is controlled. This article sets out how the report differs from the algorithm self-assessment, three preparations before drafting, the eight-chapter structure, how to arrange the evidence chain, and five frequent grounds of return.

Filing for On-Device Generative AI Services: The Boundary of Obligations and the Division of Duties between Developer and Manufacturer

On 23 September 2026 the Cyberspace Administration announced filing information for three on-device mobile generative AI services, including YOYO Claw; the announcement of 14 September showed that seven of the 124 services filed between July and August were on-device. This article explains why running on the device is not an exemption from filing, and how duties over filing, registration and public disclosure are allocated between the model developer and the device manufacturer.

Generative AI Filing Data for July and August 2026: From Market Entry to Continuous Compliance

The Cyberspace Administration's announcement of 14 September 2026 records 124 new generative AI services completing national filing between July and August (seven of them on-device), and 133 applications completing local registration; cumulative totals reached 1,112 filed services and 731 registered applications by 31 August. This article reads the data as evidence of a shift from a market-entry gate to a continuing compliance status.

Qinglang Campaign against AI Application Misuse, Phase Two: Seven Categories of Problem and Where Enforcement Is Focused

On 2 September 2026 the Central Cyberspace Affairs Commission reported the results of phase two of the Qinglang campaign against misuse of AI applications: more than 5.61 million items of unlawful information cleared, more than 49,000 accounts investigated, and over 2,400 websites and applications dealt with. This article sets out the shift of enforcement focus between the two phases, the seven categories of problem, and the points companies should check.

What Documents Large-Model Filing Requires: Two Document Sets and the Order to Prepare Them

What documents does large-model filing require? Companies that self-develop or substantially modify a model need five additional items beyond the four sets shared with algorithm filing: a security assessment report, a model service agreement, corpus labelling rules, a list of interception keywords and an assessment test question set. This article sets out the complete list for both filings, the order of preparation and time estimates.

Filing for Face-Recognition Technology: The 100,000-Person Threshold, Five Documents and the Non-Mandatory Principle

The Measures for the Security Management of the Application of Face-Recognition Technology (Order No. 19 of the Cyberspace Administration of China and the Ministry of Public Security) took effect on 1 June 2025. The Measures set the filing trigger at the storage of face information of 100,000 people, require filing with the cyberspace authority at or above the provincial level within 30 working days, and require five categories of documents. This article reviews the filing trigger, document list, the non-mandatory principle and processing rules, and their implications for AI companies.

The 'Token Economy' and Token Trading: Compliance Observations Behind AI Business-Model Innovation

In July 2026 the National Data Administration expressly encouraged business-model innovation based on token applications, exploring new trading models such as token trading, cultivating a market consensus of 'paying for quality data', and deepening the integration of token applications with industry, healthcare and finance. This article observes the contract, pricing, data and tax issues that the token economy raises from a compliance perspective.

Need legal advice on this topic?

We help foreign AI and tech companies navigate China's regulatory landscape. Get a clear, actionable assessment from bilingual lawyers who know both the law and the industry.

Request a Free Consultation →

No sales pitch. Just clear advice from lawyers who have handled cases like yours.