Skip to main content

China's Cyber Regulators Tighten Grip on AI Face-Swapping and Altered Classics

China's internet regulators are turning up the heat on AI-generated mischief. In the latest phase of the "Clear and Bright" campaign, authorities have zeroed in on everything from face-swapped videos to AI-rewritten classics, and the numbers are staggering: over 5.6 million pieces of harmful content scrubbed, nearly 49,000 accounts shown the door, and more than 2,400 websites and apps slapped with penalties. The message is clear—AI tech won't be allowed to run wild.

Local regulators have been getting creative in their approaches. Beijing nudged major platforms to sharpen their content review algorithms, making it harder for sneaky new variants of illegal content to slip through. Shanghai took a more hands-on route, pushing companies to self-audit while rolling out an "AI application knowledge package" to help the public spot red flags. Zhejiang went for a tailored touch, offering one-on-one compliance guidance to key platforms, while Guangdong kept the pressure on with regular check-ins and targeted rectifications.

Meanwhile, the big-name platforms—think Douyin, Kuaishou, Weibo, and Tencent—are beefing up their tech defenses. They're upgrading multimodal recognition models, expanding face and voiceprint databases, and rolling out real-time face governance systems. The result? Smarter detection of dodgy content, and they've published 46 governance reports to prove it.

But the crackdown isn't just about cleaning up after the fact. The giants of AI—Doubao, Yuanbao, Qianwen, and Wenxin Yiyan—are now under the microscope to tighten their content review mechanisms and slap clear labels on AI-generated material. App stores like Huawei, Xiaomi, OPPO, and vivo are also tightening their entry requirements, conducting spot checks to keep shady apps out before they ever reach your phone.

So, what's been catching the regulators' eyes? For starters, some accounts have been using AI to "remix" classics like Romance of the Three Kingdoms and Journey to the West, twisting beloved characters into sensational, low-quality plots just to chase clicks. The authorities have a colorful term for this: "digital garbage." Then there are the fake disaster videos and AI-generated rescue footage designed to stir up false alarms, not to mention AI digital humans peddling scams to the elderly.

Face-swapping and voice-cloning tech have also been abused to create inappropriate videos featuring public figures, often for profit. And it gets worse—some platforms have been flooded with bizarre, violent, or vulgar content, including disturbing animations targeting minors. There are even reports of AI "management" software being used to run fake accounts and churn out spammy posts, like digital water armies.

In response, regulators have been dishing out warnings, fines, and even shutdowns. But they're not stopping here. The cyberspace administration says it will keep a close watch on new risks as AI evolves, aiming to build a cleaner, safer online space for everyone.

Key Points

  • Massive cleanup: Over 5.6 million pieces of harmful content removed, 49,000 accounts penalized, and 2,400+ websites/apps dealt with.
  • Localized strategies: Beijing, Shanghai, Zhejiang, and Guangdong each tailored their enforcement approaches.
  • Platform upgrades: Major platforms enhanced AI detection with better models and real-time systems.
  • Source control: Leading AI models and app stores tightened review and labeling requirements.
  • Typical violations: AI-modified classics, fake disaster videos, face-swapped content, and toxic animations targeting minors.
  • Future vigilance: Regulators promise ongoing monitoring of AI-generated content risks.