What happened
In 2026, Chinese authorities intensified enforcement of AI content labelling rules, with the Cyberspace Administration of China (CAC) taking action against popular apps like Jianying and Maoxiang for non-compliance. Simultaneously, the Supreme People's Court issued unified opinions on AI disputes, establishing fault-based liability and notice-and-takedown mechanisms, while the National Financial Regulatory Administration introduced specific requirements for banks and insurers.
The Cyberspace Administration of China (CAC) moved from implementation to active enforcement of the Measures for Labelling AI-Generated and Synthetic Content in April 2026. The CAC directed local authorities to take action against the Jianying and Maoxiang apps and the Jimeng AI website for failing to properly implement labelling requirements. Actions included summoning company representatives, issuing rectification orders, and disciplining responsible personnel. This indicates that labelling compliance is now a routine supervisory focus for widely used consumer products.
On September 7, 2026, the Supreme People's Court (SPC) issued the Opinions on Lawfully Hearing Cases Involving Artificial Intelligence Disputes. The 24-article document provides a unified approach for lower courts, applying the general fault-based liability principle under the Civil Code. It establishes a notice-and-takedown mechanism for , where providers are liable if they fail to act on valid infringement notices. The opinions also address copyright, patent, and personality rights issues, emphasizing that outcomes depend on documentation of training data sources and content-filtering mechanisms.
On June 18, 2026, the National Financial Regulatory Administration (NFRA) released guiding opinions on AI in the banking and insurance sectors. The guidance covers governance structure, data governance, computing power, and risk management. It requires institutions to establish full lifecycle management systems for AI and prudently explore financial AI agents. The NFRA will supervise compliance and penalize violations, meaning foreign-invested banks and AI vendors must expect these requirements in supervisory reviews and procurement due diligence.
Source details: china-briefing.com
Why it matters
This shift from legislative drafting to active enforcement and judicial interpretation creates concrete legal risks for companies operating in China. The CAC's actions against major consumer platforms demonstrate that labelling compliance is now a routine supervisory priority, not just a theoretical obligation. For businesses, the SPC's guidance clarifies that liability is fault-based but hinges on documentation of training data and content filtering, while the NFRA's rules impose strict governance and risk management standards on financial institutions using AI. These developments require companies to update compliance frameworks, vendor contracts, and internal documentation to avoid penalties and litigation.
The CAC's enforcement against major apps signals that AI content labelling is no longer optional or theoretical; it is an active regulatory priority with tangible penalties. Companies publishing AI-generated content in China must ensure labels are present and not removed during editing, as non-compliance can lead to rectification orders and personnel discipline.
The SPC's opinions clarify the legal landscape for AI disputes, confirming that businesses are accountable for both training data and generated content. The fault-based liability standard means companies are not automatically liable for all AI output, but they must be able to demonstrate reasonable safeguards, such as content filtering and complaint handling procedures, to avoid liability in court.
For the financial sector, the NFRA's guidance introduces specific obligations, including data management and risk assessment. This creates new compliance burdens for banks and insurers, as well as their AI vendors, who must now meet stricter standards for model deployment and data security to operate in China's financial market.
Interactive Mechanism: How It Actually Works
Explore the underlying technology behind this development interactively.
crm_get_transaction(id='4092').Why can ethical evaluation not be reduced to one model score?
What to watch next
Monitor further CAC enforcement actions against other platforms, the first court cases applying the SPC's new AI dispute guidelines, and how financial institutions implement the NFRA's requirements in their procurement and risk management processes.
Watch for additional CAC enforcement actions against other platforms or apps that fail to meet AI content labelling standards, which could indicate a broader crackdown on non-compliant AI services.
Monitor the first court cases that apply the SPC's new opinions on AI disputes, as these will provide practical insights into how fault-based liability and notice-and-takedown mechanisms are interpreted in real-world litigation.
Observe how Chinese banks and insurers implement the NFRA's requirements, particularly in their vendor management and risk assessment processes, as this will set the standard for AI adoption in the financial sector.