Twelve new civil-liability articles on artificial intelligence took effect in China in late 2024, according to legal news outlet Lexology, marking the country's first dedicated framework for AI disputes.

A Beijing judge reviews a stack of AI-related civil filings. Her docket has tripled in two years. A handwritten note flags one case involving a face-swap video.

The Supreme People's Court replaced years of ad hoc reasoning with the Provisions on Several Issues Concerning the Application of Law in the Trial of Civil Cases Involving Artificial Intelligence. The framework governs provider liability, AI-generated content, training data, and personal information protection.

Lexology framed the move as a shift "from cases to rules" — from improvisation toward a uniform civil-liability system that litigants and counsel can plan around.

What does the framework cover?

The provisions run to twelve articles. Article 1 sets the scope. Article 2 defines AI service providers and AI-generated content. Articles 3 through 5 address liability for outputs that infringe third-party rights or breach contracts. Articles 6 through 8 cover training-data disputes. Articles 9 through 11 address personal information protection. Article 12 sets the effective date.

AI service providers must exercise "reasonable care" over system outputs, must label generated content where it may mislead the public, and must bear civil liability in cases where outputs cause harm to affected third parties.

Why did the SPC act now?

Chinese judges handled more than 1,800 AI civil cases between 2023 and mid-2024, per SPC statistics released in August 2024. Many involved deepfake images, voice cloning, or training-data claims that existing tort law could not resolve cleanly.

That docket forced the SPC's hand. Pre-existing contract, IP, and privacy statutes offered no clear answers in cases where algorithmic outputs caused documented harm, leaving lower courts across China to improvise by analogy.

How does it fit with earlier rules?

The framework slots into a three-tier structure that began in 2023. The Generative AI Services Regulation took effect on August 15, 2023, regulating service providers as a class. The Interim Measures for Generative AI Services followed later that same year, addressing specific compliance duties.

The SPC's late-2024 provisions complete the third tier, governing civil litigation between private parties rather than the broad administrative duties set out in the earlier regulatory instruments.

What does it mean for practitioners?

Tech-company counsel should expect more predictable outcomes on training-data and privacy claims going forward. Insurance underwriters covering AI products will study the text for exposure language.

Foreign companies with users in China now face potential civil suits in Chinese courts, while plaintiffs gain defined remedies and service providers carry documented duties supporting negligence claims by injured parties across multiple categories of dispute.

For global firms, the provisions offer the first detailed roadmap of Chinese civil-liability standards and a template to compare against the EU AI Act's liability rules taking effect across European jurisdictions on a staggered schedule through 2026.

What's next?

The provisions will be tested in lower courts through 2025. SPC judges will likely publish guiding cases as disputes multiply.

Lawyers expect further guidance on copyright in training corpora. Comparable rules abroad remain unresolved in major markets including the United States.

For now, practitioners should audit training-data provenance and review labelling policies on existing models before any new claims land in court.