China’s Supreme Court Clarifies AI Liability — Including Criminal Exposure
On September 7, 2026, China’s Supreme People’s Court issued the Opinions on Lawfully Adjudicating Cases Involving Artificial Intelligence Disputes (the “Opinions”), a wide-ranging document addressing civil disputes from AI face-swapping to model training and assisted driving. While framed as civil guidance, the document carries significant criminal law implications — especially in Articles 19 and 20 and the general principles — that could reshape how developers, providers, and users are held accountable when AI is used to commit crimes.
Key Criminal-Law Signals in a Civil Document
Noted criminal defense lawyer Shao, who specializes in new-economy cases, points to several provisions that venture beyond civil remedies. The Opinions emphasize that AI actors must respect legal boundaries and public order — language that criminal courts increasingly use to infer intent or negligence. Article 19 appears to address scenarios where AI tools are deployed for fraudulent or infringing activities, potentially imposing liability on those who knowingly enable misuse. Article 20, meanwhile, may cover duty-of-care failures — for instance, a provider that ignores obvious misuse patterns could face complicity or negligence charges.
For the crypto and Web3 sector, the implications are direct: AI-powered trading bots, automated KYC/AML tools, and decentralized identity systems all fall under China’s regulatory gaze. If an AI agent executes a fraudulent transaction or assists in money laundering, the developer who wrote the code, the provider hosting the model, and the user who deployed it could each face criminal exposure depending on their knowledge and control.
Industry Impact: Compliance Becomes a Defense
This guidance signals that courts will scrutinize whether AI actors implemented safeguards — audit trails, kill switches, misuse monitoring — as evidence of good faith. For blockchain projects integrating AI, robust governance frameworks are no longer optional; they are potential legal shields. Expect increased demand for legal review of AI model training data, output filtering, and user-access controls, particularly for cross-border operations where Chinese users may be involved.
Forward-Looking Perspective
As AI and crypto converge, regulators worldwide are wrestling with accountability. China’s move is among the first to explicitly bridge civil and criminal liability in AI disputes. For developers and providers, the message is clear: proactive risk management and transparent documentation will be critical. Shao anticipates that future cases will test the boundaries — especially where decentralized systems make it hard to pinpoint a single responsible party. The Opinions may be civil in name, but their criminal shadow will loom large over every AI deployment in China.
