Measures for Managing Generative Artificial Intelligence Services
China's regulatory framework for generative AI services, requiring service providers to adhere to core socialist values and ensure content generated by AI systems doesn't subvert state power or leak state secrets. Includes requirements for algorithm registration, data security, and user protection measures. Companies must conduct security assessments and obtain relevant administrative licenses.
| Jurisdiction | China |
|---|---|
| Category | AI Regulations |
| Status | Active |
| Latest development |
Analysis
Key Requirements
The Measures for Managing Generative Artificial Intelligence Services in China impose several key requirements on service providers:
- Security assessment and licensing: Providers must conduct security assessments and obtain relevant administrative licenses before offering generative AI services to the public.
- Content moderation: Service providers are responsible for content generated by their AI systems and must ensure it adheres to core socialist values and does not subvert state power or leak state secrets.
- Data security: Providers must use data from legitimate sources, respect intellectual property rights, and protect personal information in accordance with Chinese laws.
- Algorithm registration: Companies are required to file information about their algorithms with the Cyberspace Administration of China (CAC), including algorithm type and self-assessment reports.
- User protection: Providers must establish complaint mechanisms and handle user reports promptly.
Compliance Challenges
Organizations face several challenges in complying with these regulations:
- Balancing innovation and regulation: Companies struggle to innovate while adhering to strict content control requirements.
- Ensuring content accuracy: The requirement to prevent false information generation poses significant technical challenges for AI models.
- Data sourcing and management: Obtaining and verifying legitimate data sources while respecting intellectual property rights is complex.
- Algorithmic transparency: Providing detailed information about proprietary algorithms to regulators may raise concerns about trade secrets.
Implementation Best Practices
To implement these regulations effectively, organizations should:
- Establish a robust content moderation system with human oversight to ensure AI-generated content complies with regulations.
- Implement strong data governance practices to ensure data used for AI training is from legitimate sources and properly protected.
- Develop comprehensive documentation of AI models, algorithms, and data sources to facilitate regulatory filings and assessments.
- Regularly conduct internal security assessments and update AI systems to address potential vulnerabilities.
Recent Updates
- The Interim Measures for the Management of Generative Artificial Intelligence Services came into effect on August 15, 2023.
- In May 2024, the National Information Security Standardization Technical Committee (NISSTC) released new draft regulations for public comment until July 22, 2024.
Related Regulations
- Personal Information Protection Law (PIPL): Governs the collection and use of personal data, including in AI applications.
- Cybersecurity Law: Provides the overarching framework for data security in China.
- Provisions on the Management of Deep Synthesis Internet Information Services: Regulates deep fake technologies and synthetic media.
Industry Impact
The regulations have significant impacts on the AI industry in China:
- Market entry barriers: The licensing and assessment requirements may limit the number of companies able to offer generative AI services.
- Innovation constraints: Strict content control measures may hinder the development of certain AI applications.
- Increased compliance costs: Companies need to invest heavily in content moderation and data governance to meet regulatory requirements.
Sources
- China Releases New Draft Regulations on Generative AI
- What You Need to Know About China's New Generative AI Measures
- AI Watch: Global regulatory tracker - China
- China's New Rules For Generative AI: An Emerging Regulatory Framework
- How will China's Generative AI Regulations Shape the Future? A DigiChina Forum
- China finalizes generative AI regulation
- Basic Safety Requirements for Generative Artificial Intelligence Services
- New Generative AI Measures in China
- China's Interim measures for the Management of Generative Artificial Intelligence Services
- Translation: Personal Information Protection Law of the People's Republic of China
- China Passes New Cybersecurity Law
- Translation: Internet Information Service Deep Synthesis Management Provisions
Recent developments
- — Future of Privacy Forum provides an updated, detailed comparison between the draft and final **Interim Measures for the Management of Generative AI Services**, highlighting relaxed penalties, expanded provider obligations on training data quality, transparency requirements, and clarified scope of application, and assessing implications for compliance programs of AI service providers[2]. (source)
- — Business & Human Rights Resource Centre notes that China’s 2023 **Administrative Measures for Generative AI Services** already require AI-generated content to be labelled, and contrasts this with new Korean and upcoming EU transparency rules, illustrating how China’s regime is influencing global regulatory approaches and compliance expectations for generative AI providers[8]. (source)
- — Wikipedia’s updated entry summarizes the current status of China’s **Interim Measures for the Management of Generative AI Services**, confirming that they remain in force, outlining their institutional basis (CAC plus six other regulators), and emphasizing their focus on public-facing generative AI services and on balancing innovation with security and social stability[4]. (source)
- — Digital Watch Observatory publishes an updated resource on the **Interim Measures for the Administration of Generative AI Services**, restating their legal basis, effective date, and main obligations (data governance, content controls, cooperation with oversight inspections), and referencing how these measures fit into China’s broader digital policy framework and ongoing enforcement landscape[1]. (source)
- — China Briefing’s refreshed analysis explains how the **Generative AI Measures** apply only to services accessible to the general public in China, exclude internal enterprise and research uses, and ease earlier proposed fines and rectification timelines, noting that these changes have been welcomed by industry as providing a more workable environment for commercial generative AI deployment[3]. (source)
- — Finance Yahoo (via South China Morning Post reporting) reviews China’s rules for generative AI—based on the **Interim Measures for the Management of Generative AI Services**—emphasizing requirements to uphold core socialist values, avoid harmful or false information, respect IP and non‑discrimination in training data, and noting industry impact in the form of tightened content moderation and risk management for domestic AI platforms[6]. (source)
- — Future of Privacy Forum’s policy commentary (updated in early 2026) highlights institutional shifts in the Measures (less centrality of CAC, clearer allocation of responsibilities across agencies), new user responsibilities, and softer sanctions, and notes that these features have prompted many AI firms to adjust governance structures, documentation practices, and oversight processes to satisfy conduct‑based obligations[2]. (source)
- — Digital Watch’s 2026 update points out that under Article 19 of the Measures providers must cooperate in **oversight inspections** and disclose training data sources, scale, annotation rules, and algorithmic mechanisms, and reports that this has had a significant impact on industry by increasing demands for technical explainability and regulatory‑grade documentation in Chinese generative AI offerings[1]. (source)
- — China Briefing discusses ongoing implementation of the **Generative AI Measures**, noting that alleviated punishments and the removal of explicit fines have reduced immediate enforcement risk but still leave providers subject to rectification orders and other supervisory tools, which has led companies to invest in compliance-by-design (content filtering, data provenance checks, user verification) to avoid regulatory intervention[3]. (source)
- — Finance Yahoo reports continued application of China’s generative AI rules, underscoring obligations on providers to stop services and report illegal activities, implement safeguards against addiction for minors, and ensure “healthy” content, and notes that these requirements are shaping product design, age‑verification mechanisms, and business strategies of generative AI platforms in the Chinese market[6]. (source)
- — South Korea’s AI-labeling move is cited as part of the broader global transparency trend; the item notes that China’s 2023 generative AI rules already require AI-generated content to be marked as such, underscoring continued policy influence beyond China’s borders.[7] (source)
- — The Library of Congress analysis notes that the final rules also restricted retaining unnecessary personal information and allowed authorities to direct technical measures against non-compliant services. This indicates a strong enforcement posture alongside the innovation framework.[6] (source)
Read the full guide to Measures for Managing Generative Artificial Intelligence Services
Related regulations
- Internet Information Service Algorithmic Recommendation Management Provisions — China, Active
- National Artificial Intelligence Policy Bill — Chile, Active
- Artificial Intelligence Liability Directive — European Union, Active
- Federal Law on the Development and Use of Artificial Intelligence — Mexico, Active
- Presidential Regulation on the Framework for Artificial Intelligence Development — Indonesia, Active
- Law Establishing the Framework for the Development and Use of Artificial Intelligence — Peru, Active
- California Consumer Privacy Act AI Amendments — California, Active, effective 2020-01-01
- GPAI Framework for Responsible AI — AI Regulations, Active
Put it into practice
- Generate the policy: NIST AI RMF policy generator (generatepolicy.com)
- Buy the policy pack: NIST AI RMF Implementation Policy (cyberpolicy.shop)
- Build it yourself: Pillar 06 Companion — The 2026 AI Risk Register (ciso.diy)
Browse all regulations · Compliance deadlines · Latest updates