On 1 February 2026 the Measures for the Supervision and Administration of Live-Streaming E-Commerce took effect. Promulgated by the State Administration for Market Regulation and the Cyberspace Administration of China as Order No. 117, the measures are the first dedicated departmental rule in the live-streaming e-commerce field.
Why digital-human streamers are specifically regulated
Article 37 of the measures provides that where live-room operators use figures or videos generated by artificial intelligence and other technologies to conduct live-streaming e-commerce, they must comply with the provisions of relevant laws, regulations and rules. According to the regulator’s interpretation, the purpose of this clause is to prevent artificial intelligence from becoming a “shield” for generating or spreading disinformation.
The distinguishing feature of live-streaming e-commerce is the combination of “immediacy” and “sales”: consumers make purchase decisions within a short time based on the streamer’s image and script. If the streamer’s image is AI-generated and not labelled, consumers may mistake it for a real-person recommendation, undermining the truthfulness of the basis for their decision.
Compliance implications for practitioners
First, the labelling duty. Those using digital-human streamers should ensure viewers can identify their non-human nature. This dovetails with the labelling requirements established by the Measures for Labelling AI-Generated Synthetic Content; see enforcement of synthetic-content labelling.
Second, clarifying the responsible party. The measures continue the responsibility structure of “live-room operator — live-streaming marketer — platform”. Using an AI streamer does not transfer the operator’s responsibility; the operator remains responsible for the truthfulness and legality of the streamed content.
Third, compliance of scripts and advertising. An AI streamer promoting goods is equally bound by the Advertising Law and the Anti-Unfair Competition Law, and must not make false or misleading commercial promotions. See our commercial contracts service.
Fourth, authorisation of likeness and voice. Where a digital human’s likeness or voice imitates a real person, corresponding authorisation must be obtained, failing which it may infringe portrait rights or voice interests. See our note on the AI face-swap drama portrait-rights case and our IP protection service.
A broader signal
Live-streaming e-commerce is one of the earliest scenarios in which AI-generated content was commercially deployed at scale. Bringing digital-human streamers under regulation essentially extends the rules on “labelling generated content” from content platforms to transaction scenarios.
This extension points to a trend: compliance requirements for AI-generated content are landing scenario by scenario across industries — first the content-labelling measures, then the live-streaming e-commerce measures, with detailed rules in advertising, finance and healthcare worth watching next.
Recommendations
Merchants and MCN agencies using digital-human streamers can start with three things: check whether current live-streaming workflows include an AI-labelling step; map the authorisation sources for the digital human’s likeness and voice; and establish a sampling and archiving mechanism for AI streamer content. See our data compliance service and reporting channels for AI misuse.
A digital-human streamer will not be banned merely because it “is AI”, but it will be held to account for being “unlabelled” or “misleading”. The key to compliance is to explain both the technical capability and the boundaries of responsibility clearly.
This article is based on the Measures for the Supervision and Administration of Live-Streaming E-Commerce and related interpretation by the State Administration for Market Regulation. It is for general reference only and does not constitute legal advice.