July 29, 2026, (Inside AI) — China's cyberspace regulator has unveiled a draft law targeting cyberbullying, explicitly including harms enabled by artificial intelligence. The proposed rules, published by the Cyberspace Administration of China (CAC) on Wednesday, signal a new regulatory frontier where AI-generated content intersects with online abuse.
The draft mandates that internet platforms implement real-time monitoring systems to detect and remove bullying content, including deepfakes and AI-generated harassment. Companies failing to comply face fines up to 200,000 yuan ($29,560), with severe cases triggering business license revocations. Individuals using AI to create or spread abusive material could face criminal liability under existing laws.
This move arrives as China's internet population exceeds 1.1 billion users, where cyberbullying incidents have surged alongside generative AI adoption. The CAC specifically cites "AI-enabled abuse" as a distinct threat category, reflecting growing global alarm over deepfake harassment, voice cloning attacks, and automated hate speech campaigns. A 2025 study by the China Internet Network Information Center found that 34% of reported cyberbullying cases involved some form of synthetic media.
The draft law introduces a tiered liability framework. Platforms must establish dedicated reporting channels, respond to complaints within 24 hours, and preserve evidence for law enforcement. Repeat offenders face escalating penalties, while victims gain clearer paths to legal redress. The CAC reserves authority to order immediate content removal and suspend non-compliant services.
"This legislation closes a critical gap where existing laws failed to address the unique harms of AI-generated bullying," said Professor Li Wei, director of the Center for Internet Governance at Tsinghua University, in an interview with Inside AI. "** The challenge will be enforcement, given the speed and scale of AI content creation.**"
China's approach parallels emerging global efforts but with distinctive characteristics. The European Union's AI Act imposes transparency requirements on synthetic content, while U.S. state laws like California's AB 602 target deepfake pornography. However, China's draft goes further by mandating proactive platform monitoring rather than relying solely on user reporting, a model consistent with its broader internet governance philosophy.
Technical implementation remains uncertain. Detecting AI-generated bullying requires advanced classifiers capable of distinguishing harmful synthetic content from legitimate uses. Research from MIT and Stanford has shown that even state-of-the-art detectors struggle with adversarial examples, raising questions about the law's practical efficacy. The draft does not specify technical standards, leaving platforms to develop proprietary solutions that could fragment the enforcement landscape.
The law also addresses "cyber manhunts," a form of coordinated online harassment where groups weaponize personal information. AI tools can automate doxxing and amplify mob behavior, making these campaigns more damaging. Provisions require platforms to prevent the spread of unauthorized personal data and hold organizers accountable for resulting harm.
Industry response has been cautious. Major platforms like WeChat and Douyin have already deployed AI moderation tools, but smaller firms may struggle with compliance costs. The draft includes a grace period for implementation, though details remain vague. Critics warn that overly broad enforcement could chill legitimate speech, a recurring tension in China's regulatory environment.
The CAC is soliciting public feedback until August 28, 2026, with finalization expected by year-end. If enacted, China would become the first major jurisdiction to specifically criminalize AI-enabled cyberbullying, potentially influencing global norms. For now, the draft represents a significant escalation in the battle against weaponized AI, forcing platforms to confront harms that existing laws barely acknowledge.