China details when deepfakes, AI ‘hallucinations’ may constitute crimes

Beijing, Sep 7 (EFE).- China’s Supreme People’s Court on Monday released judicial guidelines specifying the circumstances under which the use of deepfakes and other forms of AI-generated content may incur legal liability.

According to the court, creating or distributing AI-generated replicas of a person’s face or voice without consent, as well as using such content to spread false information or damage another person’s reputation, may be subject to legal action, state news agency Xinhua reported.

The guidelines also address AI «hallucinations,» instances in which models generate fabricated or inaccurate information, one of the industry’s most significant technical challenges.

Under the new rules, AI service providers may be held legally liable if they fail to take timely action after users report content that infringes their rights or interests.

Users themselves may also face legal consequences if they deliberately prompt AI systems to generate illegal or harmful content targeting others.

The court further identified algorithmic price discrimination as an unlawful practice, referring to the use of algorithms to offer different prices or conditions for the same product or service to different consumers without a justified reason, a practice that has become widespread on some Chinese e-commerce platforms.

“We cannot expect every consumer to become an expert at spotting deception. The law must step in promptly to protect consumers’ legitimate rights and interests,» said Zhou Jiahai. head of the research office of the Supreme People’s Court.

The measures come amid rapid growth in China’s AI sector following the launch of advanced models such as Moonshot’s Kimi K3 and Zhipu’s GLM-5.2, and as Beijing and Washington compete for leadership in strategically important technologies.

Chinese President Xi Jinping has called for AI supervision and governance to be «accurate and effective» and urged authorities to continuously refine regulatory measures to prevent the technology from «getting out of control.» EFE

lcl-sk