China Interim Measures for Generative AI Services (AIGC): AI Compliance Requirements
China's Interim Measures for the Management of Generative Artificial Intelligence Services (issued by CAC and six agencies, effective August 15, 2023) applies to any organization providing generative AI services to the general public within China — including overseas companies with users in China. Requirements cover training data legitimacy, content accuracy, security assessments, algorithm registration (for services with public opinion potential), and prohibition of discriminatory or illegal content generation. Non-compliance may trigger service suspension or penalties under parallel laws (Cybersecurity Law, PIPL, Data Security Law).
Summary of publicly-available regulatory text as of 2026-08-22. Verify against current official sources before relying on this for compliance decisions. Not legal advice.
Key Facts
August 15, 2023
Service suspension; fines under Cybersecurity Law (up to ¥1,000,000) and PIPL (up to ¥50,000,000 or 5% revenue)
What Your Business Must Do
5 compliance requirements identified. Critical requirements carry the highest risk of enforcement action.
Generative AI Security Assessment & Algorithm Filing
CriticalBefore launching a generative AI service to Chinese users with 'public opinion properties or the capacity for social mobilization,' complete a security assessment and register algorithms with the Cyberspace Administration of China (CAC). Applies to chatbots, content generators, and recommendation systems with public reach.
Deadline: August 15, 2023
Interim Measures for the Management of Generative AI Services, Art. 17Training Data Legal Compliance
High PriorityEnsure all training data is from legitimate sources. Do not use personal data without consent per PIPL. Do not use copyrighted content without rights authorization. Take measures to ensure training data is truthful, accurate, objective, and diverse. Document data provenance for regulatory review.
Deadline: August 15, 2023
Interim Measures for the Management of Generative AI Services, Art. 7AI-Generated Content Labeling
High PriorityAdd visible ("explicit") labels to AI-generated text, images, audio, video, and virtual scenes, and embedded ("implicit") labels in file metadata, per the CAC/MIIT/MPS/NRTA "Measures for Labeling of AI-Generated Content" (promulgated 2025-03-14, effective 2025-09-01 — NOT the 2023 Interim Measures' general principle, which this requirement previously and incorrectly cited as the effective date; a 2-year error). Compliance is further specified by the mandatory national standard GB 45438-2025. Platforms/service providers must verify, propagate, and preserve labels/metadata, and may not tamper with labels. Ensure generated content does not include prohibited content: fake news, disinformation, pornography, or content endangering national security or public order.
Deadline: September 1, 2025
"Measures for Labeling of AI-Generated Content" (2025); GB 45438-2025User Real-Name Verification
Medium PriorityRequire users to register with real-name information before accessing generative AI services within China. Maintain records of user interactions as required for regulatory inspection. Cooperate with CAC and relevant authorities on supervision and inspection requests.
Deadline: August 15, 2023
Interim Measures for the Management of Generative AI Services, Art. 9 (incorporating the Cybersecurity Law's network real-name system, Art. 24)User Complaint and Reporting Mechanism
Medium PriorityEstablish a mechanism for users to report AI-generated content that is false, illegal, or harmful. Process complaints promptly and document resolutions. CAC may inspect complaint records during supervisory visits.
Deadline: August 15, 2023
Interim Measures for the Management of Generative AI Services, Art. 15Who Does This Apply To?
Applies to organizations providing generative AI services to the "general public within China" — this includes both Chinese-registered entities and overseas companies accessible in China (unless blocked by the Great Firewall). Key scope trigger: if your GenAI product is accessible to Chinese users and has "public opinion properties" (influencing public discourse) or "social mobilization capability," you must complete CAC security assessment and algorithm registration. B2B enterprise AI tools with controlled access (whitelisted users, VPN-required) may have lighter obligations. Non-public research/development uses are excluded.
Recent Enforcement Actions
Against: 灵象智问AI / Lingxiang Zhiwen AI (operator: Chongqing Lingxiang Zhiwen Technology Co., Ltd.); 重庆哨兵拓展迷; 南川区蓉城网络科技工作室 (among four providers)
Chongqing's internet regulator penalized four AI service providers under the Interim Measures for Generative AI Services and the Cybersecurity Law — three for offering public-facing generative AI services without completing the required security assessment and large-language-model algorithm filing (灵象智问AI/Lingxiang Zhiwen AI, 重庆哨兵拓展迷, and 南川区蓉城网络科技工作室's unauthorized ChatGPT-based service), and one (开山猴 AI writing site) for disseminating illegal content due to inadequate content moderation. The three unauthorized-service operators were summoned and ordered to suspend the relevant services pending rectification. Separately, the national CAC published its first list of filed generative-AI services on 2 Apr 2024, with 302 services completing national-level filing by end-2024.
SourceRecent Regulatory Guidance
CAC + MIIT + MPS + NRTA — Measures for Labeling of AI-Generated Content (effective 2025-09-01)
Four regulators jointly promulgated the Measures for Labeling of AI-Generated Content on 2025-03-14, effective 2025-09-01: explicit (visible) labels required on AI-generated text, images, audio, video, and virtual scenes; implicit labels required in file metadata; platforms/service providers must verify, propagate, and preserve labels; tampering with labels is prohibited. Implementation detailed by mandatory national standard GB 45438-2025. This is a DIFFERENT, later, more detailed instrument than the 2023 Interim Measures' general labeling principle.
SourceCAC national generative-AI service filing list (first published)
CAC published its first national list of algorithm-filed/registered generative AI services on 2024-04-02, with 302 services completing national-level filing by end-2024 (per the White & Case-sourced Chongqing enforcement action above). Registration/filing volume has continued growing — a 2025 update reported dozens more newly filed and registered services in a single reporting round (Beijing/Shanghai). No independently confirmed CAC "outreach to OpenAI/Anthropic/Google" exists; in practice, overseas GenAI providers respond to China's regulatory perimeter by geofencing (foreign-phone-number/card requirements) rather than direct CAC compliance outreach.
SourceQuarterly Enforcement Digest
Q3 2026 update: CAC actively enforcing AIGC Interim Measures domestically — the named 2024-07-22 Chongqing action (four providers, service suspensions) remains the clearest documented enforcement example. The mandatory AI-content-labeling regime (Measures for Labeling of AI-Generated Content, promulgated 2025-03-14) took effect 2025-09-01 — a real, significant compliance deadline this entry previously mis-cited as already effective since 2023. Cycle 8 (2026-08-22) correction: removed a fabricated "CAC outreach to OpenAI/Anthropic/Google" claim with no independent corroboration — in practice, overseas providers respond to China's regulatory perimeter through geofencing (foreign-phone-number/card requirements) rather than direct CAC compliance outreach; enforcement against inaccessible overseas services is primarily via App Store removals and payment platform blocking, not fines. Domestic Chinese AI companies (Baidu Ernie, Alibaba Tongyi, Huawei PanGu) are all AIGC-registered. Foreign companies operating in China or offering Chinese-language AI services must engage local legal counsel for the CAC security assessment process AND the newer 2025-09-01 labeling requirements.
Digest covers enforcement actions, guidance, and regulatory developments. Last verified: 2026-08-22.
Industry Playbooks covering China Interim Measures for Generative AI Services (AIGC)
These industry playbooks include jurisdiction-specific checklist items and guidance for China Interim Measures for Generative AI Services (AIGC).
Frequently Asked Questions
Does China Interim Measures for Generative AI Services (AIGC) apply to my business?
China's Interim Measures for the Management of Generative Artificial Intelligence Services (issued by CAC and six agencies, effective August 15, 2023) applies to any organization providing generative AI services to the general public within China —… Use Aegis Firma's free scanner to get a personalized assessment in under 5 minutes.
What is the penalty for non-compliance?
The maximum penalty under China Interim Measures for Generative AI Services (AIGC) is: Service suspension; fines under Cybersecurity Law (up to ¥1,000,000) and PIPL (up to ¥50,000,000 or 5% revenue). Fines are typically scaled by company size, severity of violation, and whether violations were willful or accidental.
How do I comply with China Interim Measures for Generative AI Services (AIGC)?
The 5 requirements above cover the core obligations. The fastest path to compliance is: (1) conduct an AI risk assessment, (2) document your AI systems, (3) implement transparency disclosures where required. Aegis Firma generates all required documents automatically.
Official Source
https://www.chinalawtranslate.com/en/generative-ai-interim/Last updated: 2026-08-22 — verify at source before relying on this information.
Don't leave compliance to chance
Aegis Firma scans your AI tools, tells you exactly which regulations apply, and generates all required documents — in 30 minutes.
Start your free compliance scan