China’s internet regulator has published draft rules to combat cyberbullying, and, according to Reuters, the measures are written to reach abuse generated with artificial intelligence.
The Cyberspace Administration of China released the draft for public comment on 29 July, part of a widening effort to police what happens on the country’s platforms.
The draft targets what Chinese regulators call “cyberviolence.” That covers the online spread of harassing, defamatory, or false information that harms a person, a category the regulator defines broadly enough to fold in AI misuse alongside ordinary trolling and pile-ons.
Most of the weight falls on platforms. Under the draft, services would have to build classifiers that detect cyberviolence, offer users a “one-click protection” that blocks incoming messages and comments, and act on those signals by restricting abusive content and investigating the accounts behind it.
The enforcement machinery is detailed. Platforms would keep blacklists of repeat offenders, file regular reports on their interventions, and run early-warning systems to catch pile-ons before they escalate, with penalties running up to the closure of an abuser’s account.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!
Some platforms are already moving ahead of the text. The question-and-answer site Zhihu, anticipating the rules, has moved to curb anonymous accounts, a sign of how quickly Chinese companies adjust once a regulatory direction becomes clear.
The AI dimension is what makes it timely. Reuters reports that the rules are meant to cover AI-enabled abuse, the deepfaked images and synthetic text that can be produced at volume and aimed at a single target, a form of harassment that older moderation systems were never built to catch.
It fits a pattern of Chinese AI rulemaking. Beijing already requires AI-generated content to be labelled and has moved to regulate human-like AI services and a digital-human industry in which people even rent out their faces to AI, building a lattice of rules that treats synthetic media as something to be tracked rather than left alone.
The state’s interest is not only protective. The same tools that detect abuse also identify speakers, and a regime that requires platforms to log, trace, and blacklist users hands the authorities a fuller map of who is saying what, a point critics of China’s internet governance make repeatedly.
That tension runs through the draft. Framed as a shield for victims of harassment, it also deepens the obligations that make Chinese platforms answerable to the state, and the line between protecting users and surveilling them is, as ever, drawn by the government.
The victims it names are real enough. Cyberbullying campaigns in China have driven people offline and, in some cases, to worse, and public anger after several high-profile cases has given the authorities a popular mandate to act.
Anonymity sits at the centre of the fight. Much online abuse in China is posted from anonymous or pseudonymous accounts, and the draft’s logging and blacklist requirements press platforms to know who is behind them, which is exactly what makes the rules effective and what makes them worrying.
Other governments are circling the same problem. From deepfake abuse to synthetic harassment, regulators from Europe to the United States are trying to write rules for AI-generated cruelty, and China’s draft is among the more prescriptive attempts so far.
What happens next is procedural. The draft is open for public comment before it can be finalised, and the details that matter, how aggressively platforms must scan and how precisely the AI provisions are worded, may shift before the rules take effect.
The direction, though, is set. China is building the plumbing to detect and act on AI-generated abuse at scale, and whatever the final text says, the expectation it sets for platforms is that they will see everything, and answer for it.
