China is formulating regulations on cyberbullying that address abuse generated by AI.

      China's internet regulatory body has released draft regulations aimed at addressing cyberbullying, as reported by Reuters, with particular attention to abuse facilitated by artificial intelligence. The Cyberspace Administration of China made the draft available for public input on July 29, as part of a broader initiative to monitor activities on the nation’s online platforms.

      The draft addresses what Chinese authorities classify as "cyberviolence," which includes the digital dissemination of harassing, defamatory, or misleading information that harms individuals. This definition is broad enough to encompass both the misuse of AI and typical online trolling.

      The responsibility primarily falls on the platforms. According to the draft, these services would need to implement systems to identify cyberviolence, provide users with a “one-click protection” feature to block harmful messages and comments, and take necessary actions by limiting abusive content and probing into the accounts responsible for it.

      The enforcement framework is comprehensive. Platforms are expected to maintain blacklists of repeat offenders, submit regular intervention reports, and establish early-warning mechanisms to intercept bullying incidents before they escalate, with potential consequences including the suspension of abusive accounts.

      Some platforms have already started proactive measures. The question-and-answer platform Zhihu, foreseeing the new rules, has taken steps to limit anonymous accounts, reflecting how swiftly Chinese companies adapt once regulatory guidelines become evident.

      The inclusion of AI reflects the timeliness of these regulations. Reuters notes that the rules are designed to address AI-driven abuse, such as deepfakes and synthetic text that can be generated en masse and targeted at specific individuals, a form of harassment that previous moderation systems struggle to manage.

      This approach aligns with a broader trend in China's AI regulation. Authorities already require that AI-generated content be labeled and have begun regulating human-like AI services and the digital-human industry, which involves people renting out their likenesses to AI, creating a complex framework that treats synthetic media as something to be monitored rather than ignored.

      The state's motivations are not solely protective. The same technologies that detect abuse can also identify users, and a framework that mandates platforms to log, trace, and blacklist users provides the government with a more comprehensive understanding of online speech—a common concern raised by critics of China’s internet governance.

      This tension is evident throughout the draft. While it is presented as a protective measure for victims of harassment, it also increases the responsibilities of Chinese platforms to the state, blurring the line between user protection and surveillance, a demarcation consistently drawn by the government.

      The victims referenced in the regulations are indeed real. Instances of cyberbullying in China have resulted in individuals withdrawing from online spaces and, in some cases, facing dire consequences. Public outrage following several well-publicized incidents has given the authorities a popular impetus to act.

      Anonymity is a central issue in this struggle. Much of the online abuse in China originates from anonymous or pseudonymous accounts, and the draft's requirements for logging and blacklisting compel platforms to identify the individuals behind these accounts, which contributes both to the rules’ effectiveness and to the concerns surrounding them.

      Other governments are facing similar challenges. From deepfake misuse to synthetic harassment, regulators in Europe and the United States are working to create regulations addressing AI-generated misconduct, and China’s draft represents one of the more prescriptive efforts thus far.

      The next steps involve procedural actions. The draft is currently open for public commentary before finalization, and critical details—such as the extent of platform scanning required and the specificity of the AI provisions—may change prior to implementation.

      However, the intended direction is clear. China is establishing a system to detect and respond to AI-generated abuse on a large scale, and irrespective of the final wording, platforms are expected to monitor and address all instances of abuse.