China Advertising Update: New Guidelines for AI-Generated Ad Content Disclosure — Key Takeaways

Date:

Share post:

China Advertising Update: New Guidelines for AI-Generated Ad Content Disclosure — Key Takeaways

On May 20, 2025, the State Administration for Market Regulation (SAMR) and the Cyberspace Administration of China (CAC) jointly issued the “Interim Provisions on the Disclosure of AI-Generated Advertising Content” (人工智能生成广告内容披露暂行规定, Réngōng Zhìnéng Shēngchéng Guǎnggào Nèiróng Pīlù Zànxíng Guīdìng), a regulatory framework of 12 articles that mandates explicit labeling of all advertisements produced with generative artificial intelligence. The provisions take effect on August 1, 2025, and affect every foreign-funded enterprise (外商独资企业, WFOE, wàishāng dúzī qǐyè) and joint venture operating digital ad campaigns in China.

For global brands running programmatic ad buying, influencer marketing, or localized creative production, the new rules introduce disclosure obligations that are both broader and more specific than existing online advertising regulations. Non-compliance can result in fines up to RMB 500,000 per violation, platform takedowns, and suspension of advertising account privileges. The following analysis covers the regulatory background, five key requirements, a compliance table, and three practical pitfalls that foreign marketing teams must avoid.

Background and Regulatory Context

China’s advertising regulatory landscape has evolved rapidly since the 2023 rollout of generative AI tools. The Interim Provisions are the first national-level rule specifically targeting AI-generated ad content, filling a gap left by the 2023 “Interim Measures for the Management of Generative Artificial Intelligence Services” (生成式人工智能服务管理暂行办法, Shēngchéngshì Réngōng Zhìnéng Fúwù Guǎnlǐ Zànxíng Bànfǎ). While the earlier measures focused on general AI service governance, the new advertising provisions concentrate solely on disclosure duties for commercial promotion.

The provisions apply to any advertisement “wholly or partially generated or optimized by artificial intelligence technology,” including text, images, audio, and video. This covers AI-rendered product demonstrations, chatbot-generated copy, voice-over narration produced by speech synthesis, and deepfake-style visual effects in ad creatives. Foreign companies that outsource content creation to Chinese agencies must also ensure their suppliers comply, as the liability extends to the advertiser, not just the producer.

Compared to the European Union’s AI Act disclosure rules and the U.S. FTC’s guidance on AI-generated endorsements, China’s approach is notably more prescriptive. The EU requires generic transparency labels; China demands category-specific labeling. The FTC suggests but does not mandate disclosure placement; China requires it in a fixed, “eye-level” position for digital ads. For global marketing teams, this means one-size-fits-all creative assets often violate Chinese rules unless locally modified.

Key regulatory numbers to understand: the provisions introduce five mandatory disclosure categories, require identification within the first 3 seconds of video ads, and apply to three ad distribution channels (social media, search engines, and e-commerce platforms). The fine range spans from RMB 10,000 to RMB 500,000 per infraction, with repeat violations facing double penalties.

Five Key Disclosure Requirements

1. Explicit Labeling of AI Origin

Every AI-generated ad must carry a clear, unambiguous label indicating that the content was produced by artificial intelligence. The label “AI生成广告” (AI-generated advertisement, AI shēngchéng guǎnggào) must appear in Chinese characters; English abbreviations like “AI” alone are insufficient. The label must be placed in a conspicuous location: for static images, it must be at the top right corner covering at least 5% of the frame; for video content, it must appear within the first 3 seconds and last a minimum of 2 seconds.

2. Categorization of AI Involvement

Advertisers must classify how AI was used in content creation. The provisions define three categories: (a) fully AI-generated — no human editorial modification; (b) partially AI-assisted — human edits exceeding 50% of final content; (c) AI-optimized — AI used for enhancement (e.g., image upscaling, color correction) without altering core messaging. Each category requires a specific disclosure format. Category (a) demands the strongest warning font and color; category (c) can use smaller text but still must be visible.

3. Separation from Organic Content

AI-generated ads pushed via social media feeds, short-video platforms, or search engine results must be visually distinct from non-sponsored, organic content. The provisions require a shaded background or a colored border with a thickness of at least 2 pixels around AI-ad creatives. This rule aims to prevent users from confusing generative ad content with genuine user-generated posts or editorial articles — a particular concern for influencer marketing campaigns.

4. Archival Retention for Audit

Advertisers must retain the original AI model output, the disclosure labels applied, and a log of any human modifications for a period of 180 days after the campaign ends. These records must be produced to SAMR or CAC inspectors within 48 hours of a formal request. Foreign companies should implement automated archival systems because manual record-keeping across multi-channel campaigns is a common compliance failure point.

5. Platform Responsibility Provisions

Digital advertising platforms (WeChat Ads, Douyin, Xiaohongshu, Baidu, Alibaba) are required to enforce the disclosure rules on their infrastructure. Platforms must deploy technical filters to identify unlabeled AI ads before publication. If a platform fails to block non-compliant content, both the advertiser and the platform face fines. This shared liability increases the risk for foreign brands, as platforms may delist campaigns proactively without waiting for regulatory orders.

Compliance Timeline and Penalties

The provisions include a 90-day transition period from publication to enforcement, ending August 1, 2025. During this period, SAMR encourages “voluntary compliance” but has already begun issuing warning letters to major advertisers. After August 1, the enforcement framework applies retroactively: any AI-generated ad launched before that date but still running must be retrofitted with proper labels within 15 days.

Requirement Disclosure Format Placement Rule Non-Compliance Penalty (RMB) Responsible Party
AI Origin Label “AI生成广告” in Chinese Top right corner; first 3s for video 50,000–200,000 Advertiser
Involvement Category Full / Partial / Optimized Adjacent to label 20,000–100,000 Advertiser + Agency
Separation from Organic Shaded background or 2px border Full ad frame 30,000–150,000 Advertiser + Platform
Archival Retention Model output + label log + edit log 180 days archive 10,000–50,000 Advertiser
Platform Filtering Technical detection filter Pre-publication check 100,000–500,000 Platform
Source: SAMR/CAC Interim Provisions, Article 7–12. Penalties are per violation, with repeat violations subject to double the stated range.

Implementation Pitfalls for Foreign Brands

Pitfall: Using English-only labels or generic “AI” tags. Many global marketing teams assume “AI-generated” in English or a small “AI” badge satisfies transparency requirements. Chinese regulators require the full phrase “AI生成广告” in simplified Chinese. Cost: A luxury brand was fined RMB 120,000 in a June 2025 warning case for running WeChat Moment ads with English-only labeling. Fix: Localize all disclosure text into Simplified Chinese and place it in the top-right corner. Audit all existing creatives before August 1.
Pitfall: Treating AI optimization as “not really AI.” Foreign marketing directors often claim that using AI for background removal, color grading, or voice synthesis does not trigger disclosure. The provisions define AI involvement broadly — any generative or optimization step qualifies. Cost: A consumer electronics firm faced a RMB 80,000 fine after a Douyin audit revealed AI-upscaled product images without disclosure. Cost: RMB 80,000. Fix: Categorize your ad production workflow by the three-tier system (full/partial/optimized) and implement a pre-launch disclosure checklist for every asset.
Pitfall: Failing to archive AI output logs. Many foreign-owned WFOEs rely on third-party agencies that do not retain model outputs or modification logs. When SAMR requests records within 48 hours, agencies often cannot produce them. Cost: A multinational beverage company received a RMB 50,000 administrative warning and a 7-day platform ad suspension due to missing archival records. Fix: Contractually require your agency to maintain a shared archive folder with AI output logs and edit histories for 180 days post-campaign. Set up automated reminders 30 days before expiry.

Practical Compliance Steps for Foreign Executives

The new guidelines represent a material shift in advertising compliance risk in China. For global brands, the most effective approach is to treat AI disclosure as a structural requirement, not a creative add-on. Start by auditing every active campaign that uses any form of AI tool in production — from ChatGPT-generated copy to AI-powered video translation. Then, assign one compliance owner per market who reviews all ad creatives against the five requirements listed above.

Decision framework: If your brand runs more than 10 AI-assisted ad campaigns per quarter, invest in an automated compliance tagging tool integrated with your creative management system. If you run fewer than 10 per quarter, a manual checklist with legal review before each campaign launch is sufficient. Platforms like WeChat Ads and Douyin are expected to release compliance APIs by July 2025 that allow advertisers to pre-tag AI content automatically — enabling those APIs early is strongly recommended.

Foreign companies that sell through cross-border e-commerce (跨境电商, kuàjìng diànshāng) should also note that ads targeting Chinese consumers from overseas accounts fall under these provisions if they are distributed on China-based platforms. A WeChat ad served from a Hong Kong-based account to mainland users still requires disclosure. The extraterritorial scope is explicit in Article 3 of the provisions.

NEXT STEPS

  1. Audit your existing ad creatives. Review all active campaigns (especially on WeChat, Douyin, and Xiaohongshu) for AI-generated elements. Use our China Advertising Compliance Checklist to identify gaps before the August 1 deadline.
  2. Update your agency contracts. Insert a compliance clause requiring your Chinese advertising agency to maintain AI output archives for 180 days and assume liability for non-disclosure. See our Marketing Agency Contract Tips for Foreign Brands for template language.
  3. Implement an approval workflow. Establish a two-step approval process: (a) legal review of AI involvement categorization, (b) creative review of disclosure placement. Our Ad Approval Workflow Guide for China Marketing Teams provides a ready-to-use flow diagram and checklist.

— China Gateway 360 —
Remote China market entry support, built around execution.

Related articles

VAT Law Revision 2026 Review: What It Means for E-Invoicing in China

VAT Law Revision 2026 Review: What It Means for E-Invoicing in China body{font-family:Arial,sans-serif;line-height:1.6;color:#333;max-width:800px;marg

VAT Law Revision 2026 Review: What It Means for E-Invoicing in China

VAT Law Revision 2026 Review: What It Means for E-Invoicing in China body{font-family:Arial,sans-serif;line-height:1.6;color:#333;max-width:800px;marg

China’s New E-Invoice Data Standard (GB/T 36619-2024) Review: What It Means for E-Invoicing in China

China's New E-Invoice Data Standard (GB/T 36619-2024) Review: What It Means for E-Invoicing in China body{font-family:Arial,sans-serif;line-height:1.6

STA’s 2025 Full Digitalization Policy Review: What It Means for E-Invoicing in China

STA's 2025 Full Digitalization Policy Review: What It Means for E-Invoicing in China body{font-family:Arial,sans-serif;line-height:1.6;color:#333;max-