In late August 2026, Chinese regulators and media converged on one message: AI-generated avatars and digital humans are not a free pass around advertising law. For overseas brands using AI avatars to sell into China, the rules just got sharper — and the liability now lands on the advertiser, not the algorithm.
⚖️ AI avatars cannot legally endorse products
Under Article 2 of China's Advertising Law, an ad endorser can only be a natural person, a legal entity, or another organization acting on behalf of the advertiser. An AI-native avatar is an algorithm-generated digital image. It has no legal-personality status, so it cannot serve as an endorser in the legal sense.
That closes a loophole some brands were testing. A virtual presenter cannot "recommend" a product the way a human spokesperson can. Regulators have been explicit: a brand cannot use an AI avatar to sidestep endorser oversight.
🚫 The five red lines advertisers keep crossing
Beyond the endorser question, regulators listed the violations that trigger enforcement:
- Fabricated first-person experience. An avatar claiming "I wore this all day and it felt great" is a false testimonial when no human wore it.
- Absolute claims. Words like "best," "100% effective," or "guaranteed results" are banned under advertising law, whether a human or an avatar says them.
- Restricted-category promotion. Using an avatar to advertise medical, financial, or other controlled goods without the required pre-approval.
- Missing AI labels. Content that is AI-generated but not visibly marked as such.
- Skipping ad review. Failing to run the required review process before publishing.
The common thread: the advertiser is the first responsible party. "The algorithm did it" is not a defense — media commentary was direct that brands must review frame by frame and line by line.
🌍 Why overseas brands get hit harder
Three factors put overseas advertisers at a disadvantage:
- No local entity, no local compliance owner. When a violation surfaces, a foreign brand has no China-licensed body to absorb the penalty or respond to regulators directly.
- Unfamiliar red lines. Teams outside China rarely know that "best" and "100% effective" are absolute claims, or that a first-person AI testimonial is treated as false advertising.
- AI tools generate non-compliant copy automatically. A generative tool asked to write Chinese ad copy will happily produce "彻底解决" or "保证见效" phrasing that is illegal on arrival. An overseas team with no native reviewer cannot catch it.
⚠️ The compliance gap no one owns
The rules also expose a gap regulators themselves acknowledge. AI-labeling requirements remain high-level principles without granular technical standards. Network penalties mostly target the AI tool provider, not the merchant; an unlabeled ad by an ordinary merchant is usually handled as misleading advertising under market regulation, with enforcement varying by region. There is still no dedicated clause governing avatar-based promotion, so brands face inconsistent enforcement.
For an overseas brand, "inconsistent" is worse than "strict." Strict is predictable. Inconsistent means a campaign that clears in one province can be pulled in another.
🛡️ How BPP keeps your AI marketing compliant
BPP runs Baidu advertising and GEO content for international brands under a compliant structure — no China business license required on your side. For AI-avatar and AI-generated content, that means:
- Real-person endorsement. We route spokesperson and testimonial work through verified human endorsers, so you never depend on an AI avatar in the legal endorser role.
- Label-first production. Every AI-generated asset carries explicit and implicit labels before it ships.
- Copy review against advertising law. We screen Chinese-language copy for absolute claims, fabricated experience, and restricted-category violations before it goes live.
- A local compliance owner. Our China-side structure is the accountable party, so enforcement lands somewhere that can actually respond.
Key takeaways
- An AI avatar has no legal-personality status, so it cannot serve as an ad endorser under Article 2 of China's Advertising Law.
- Five red lines trigger enforcement: fabricated experience, absolute claims, restricted-category promotion, missing AI labels, and skipped review.
- Overseas brands face three extra risks: no local entity, unfamiliar red lines, and AI tools that auto-generate illegal copy.
- BPP runs your Baidu ads and GEO content under a compliant structure — real-person endorsement, label-first production, and ad-law copy review.
✅ A short checklist before you launch
If you plan AI-avatar or AI-generated marketing in China:
- Treat the avatar as a presenter, never an endorser.
- Ban "best," "100%," "guaranteed," and any first-person experience claim from AI copy.
- Label every AI-generated asset before publishing.
- Route controlled-category products through human review.
- Put a China-licensed compliance owner on paper before your first ad runs.
AI avatars cut production cost. They do not cut liability. The brands that treat the avatar as a legal shortcut will be the first ones pulled from the results — and the first ones fined.