China’s Cyberspace Administration (CAC) finalized its Measures for Labeling Artificial Intelligence-Generated Content in March 2025. In 2026, enforcement is active. Any AI-generated image, video, or audio used in advertising must carry a visible “AI??” label and an embedded watermark in the file metadata. The advertiser carries full legal responsibility. Norton Rose Fulbright and Harris Sliwoski have both published detailed breakdowns of the rule set.
The rule applies to all platforms: Douyin, WeChat Moments ads, Baidu display, Xiaohongshu sponsored posts. The SAMR’s 2026 national advertising work plan now classifies AI-generated ad content as “self-generated content,” which shifts compliance liability entirely to the brand or its appointed agency.
What Changes for Brands Using AI Creative Tools
Teams producing creatives with Midjourney, DALL-E, Sora, or Chinese tools such as Tongyi Wanxiang, Vidu, or Jianying now face two concrete obligations.
First, the visible label. Every AI-generated creative served to a Chinese consumer must display “AI??” in a legible position. No exception for cases where confusion is unlikely.
Second, the technical watermark. Each file must embed the brand’s social credit code plus a unique content identifier in its metadata. This is the harder requirement. A foreign brand without a Chinese legal entity has no social credit code. It cannot technically comply with this part of the regulation on its own. It depends entirely on a local partner or registered entity to carry the watermark.
This creates a real gap for international brands running campaigns from headquarters. A marketing team in Paris or New York that generates creatives with a Western AI tool and sends the files to a Chinese media buyer is not compliant. The watermark cannot be added retroactively by the buyer without the brand’s registered identifier.
Chinese agencies that already operate certified AI production pipelines, and that hold their clients’ social credit codes on file, can embed the watermark at generation time. Foreign agencies cannot do this unless they work through a local entity. This is a structural advantage for domestic production shops, not a regulatory side effect.
One area remains unclear: creatives produced outside China but distributed inside China. The regulation covers distribution on Chinese platforms, but enforcement against the file’s origin country is technically ambiguous. Regulators have not published specific guidance on cross-border production chains. In practice, the platform bears secondary liability if it serves non-compliant content, so platforms are already pressuring advertisers to certify compliance before campaign launch.
Who Is Exposed
Any brand sending AI-generated creatives to a Chinese media buyer for Baidu Ads, Douyin, or WeChat campaigns is exposed. This includes brands that outsource creative production to their global agency without specifying AIGC compliance in the brief.
Before launching a campaign, three things need verification:
- Does the creative contain any AI-generated element, even partial (background, character, voice-over)?
- Is the “AI??” label visible in the final ad unit?
- Does the file metadata contain the brand’s social credit code and a unique content ID?
If any answer is no, the campaign should not go live. Fines under the advertising law can reach three times the ad spend for the non-compliant period.
Building a Compliant Creative Pipeline
The simplest compliant setup has four steps.
Step 1: Confirm your production tool is certified. Chinese AI tools approved under the CAC’s generative AI service filing system (Tongyi Wanxiang, Vidu, Jianying, Kuaishou’s Kling) can embed the required watermark automatically. Western tools cannot. If your agency uses Midjourney or DALL-E, ask explicitly how they handle AIGC watermarking for China distribution.
Step 2: Add an AIGC compliance clause to your creative brief. The brief should require the agency to confirm: which AI tools were used, whether the “AI??” label appears in the final asset, and whether the metadata watermark has been applied using the brand’s registered social credit code.
Step 3: Request watermark confirmation before delivery. Ask your agency to provide a screenshot or file metadata extract showing the embedded watermark. This creates a paper trail if a platform or regulator raises a question post-launch.
Step 4: Review your media buying contracts. Liability language in Chinese media contracts often places compliance responsibility on the advertiser. Make sure your contract with the media buyer specifies who holds responsibility for AIGC compliance, and that it is not assumed to be the platform.
Brands that need a full audit of their China advertising setup can start with GMA’s China advertising services, which covers platform selection, creative compliance, and media buying across Douyin, WeChat, and Baidu.
The Debate
The AIGC labeling rules create a structural difference between domestic Chinese agencies and international ones. Local agencies have the social credit codes, the certified tools, and the platform relationships. Is this regulation a genuine consumer protection measure, or does it function as a barrier that pushes international brands to use Chinese production partners for every campaign?
