China’s Cyberspace Administration (CAC) ordered ByteDance’s video editing apps Jianying and Maoxiang and its AI website Jimeng AI to comply with rules on labeling AI-generated content, the regulator said in a statement on April 28, according to Reuters. The apps failed to adequately implement required AI content identification measures.
From Rules to Enforcement
China implemented AI content labeling requirements in 2024, mandating that platforms clearly identify AI-generated text, images, audio, and video. For nearly two years, enforcement remained largely at the policy-setting level. The ByteDance action marks a visible shift to direct regulatory intervention against China’s largest tech companies.
This is not an isolated enforcement action. In February 2026, the CAC announced that over 13,400 non-compliant accounts were penalized and over 543,000 pieces of illegal and non-compliant information were removed across platforms, according to an Oxford Global Society analysis. The scale of removals indicates systematic monitoring rather than selective enforcement.
The Agent Generation Problem
The compliance burden has specific implications for AI agent deployments in China. As agents generate content autonomously at scale (marketing copy, customer communications, social media posts, video editing outputs), every piece of agent-generated content potentially falls under the labeling requirement.
For platforms like Jianying (a video editing app with AI features), the challenge is tracking which portions of output were AI-generated when the tool blends human editing with AI assistance. The line between “AI-assisted” and “AI-generated” becomes ambiguous in tools where users provide creative direction but AI executes specific tasks.
ByteDance is China’s most important test case because of its scale. TikTok’s parent company operates multiple AI-powered creative tools serving hundreds of millions of users in mainland China alone. If the CAC enforces labeling compliance against ByteDance successfully, smaller platforms face implicit pressure to self-correct.
Regulatory Divergence
The enforcement action arrives one day after China’s NDRC blocked Meta’s $2 billion acquisition of agentic AI startup Manus on national interest grounds. Together, the two actions illustrate China’s dual approach: protecting domestic AI companies from foreign acquisition while tightening compliance requirements on their behavior.
The contrast with Western approaches is notable. The EU AI Act requires AI content labeling but enforcement mechanisms remain under development. The U.S. has no equivalent federal requirement. China is moving faster on enforcement while simultaneously using regulatory power to shape the competitive landscape for AI companies operating within its borders, according to Brussels Morning.
For international enterprises deploying AI agents in China, the message is operational: AI content labeling is not a future requirement to plan for. It is an active enforcement priority with demonstrated penalties for non-compliance.