Here is a .com classroom post version, corrected and shaped for reading, discussion, and a student field report.
China’s New Rules for Digital Humans — and What the West Should Notice
On April 3, 2026, China’s Cyberspace Administration released draft rules for “digital virtual humans,” with public comments open until May 6, 2026. The draft targets online services that use virtual human figures to simulate appearance, voice, behavior, interaction, and even personality. It covers both AI-generated and human-driven versions, such as motion-capture performers. (European Commission)
This matters because the draft does not only regulate the AI model in the background. It regulates the character in front of the user — the digital person, avatar, influencer, companion, or assistant that people meet on the screen. That is a major shift in AI governance. (European Commission)
What China is actually saying
The draft says that if a digital human appears in a public-facing online service, it must be clearly labeled as a digital human from the very beginning and throughout the service. This is not just hidden metadata. It is a continuous, visible notice meant to prevent confusion between real people and synthetic personas. (European Commission)
The draft also says that if a company uses a real person’s sensitive personal information to build such a figure, it must obtain clear and separate informed consent. If the person withdraws consent, the relevant data must generally be deleted and the digital human canceled unless the parties agreed otherwise under the law. For children under 14, guardian consent is required. (European Commission)
There is another important protection: providers may not create or use digital humans with identifiable traits of real people without consent. The rule reaches beyond obvious face-swaps. It includes things like highly similar voices, portraits, pen names, stage names, screen names, and other recognizable identity markers. (European Commission)
China’s draft is also unusually explicit about minors. It prohibits services that push minors toward addiction and bars certain relationship simulations, including virtual family-type or intimate relationships aimed at minors. It also prohibits inducing excessive spending and content that could harm minors psychologically or physically. These protections apply to minors broadly, while the stricter consent rule above is specifically for children under 14. (European Commission)
The draft also contains broad content restrictions tied to national security, public order, and what China calls Core Socialist Values. It bans content involving terrorism, separatism, extremism, obscenity, gambling, violence, rumors, defamation, and similar categories. It also forbids using digital humans to bypass identity checks such as face or voice authentication. (European Commission)
One especially revealing part is that the draft does not treat digital humans as neutral entertainment. It treats them as potentially manipulative interactive systems. It says users who try to exit or cancel a service must not be tricked or excessively induced to stay, and it calls for intervention where users show signs of self-harm risk. (European Commission)
What makes this interesting outside China
Western countries are already discussing AI transparency, deepfakes, impersonation, and harmful synthetic media. The EU AI Act is now in force and includes transparency obligations for certain AI uses, including deepfakes, as part of a broader legal framework for trustworthy AI. (European Commission)
In the United States, the FTC has been warning about voice cloning harms, impersonation scams, and the limits of self-regulation. It has pursued consumer-protection approaches and highlighted risks from AI-enabled voice imitation and fraud. (Federal Trade Commission)
But China’s draft goes a step further in one specific way: it focuses directly on the interaction layer. Not just the model. Not just the training data. Not just watermarking. It focuses on the synthetic person that speaks, persuades, flirts, comforts, sells, teaches, or influences. (European Commission)
That is what Western regulators may need to study more carefully.
What the West could learn — without copying China’s politics
The first lesson is simple: visible labeling matters. If a system is a digital human, users should know that immediately and continuously. A persistent label is easier to understand than a buried policy page or invisible watermark. China’s approach is stricter and more user-facing than many current Western discussions. (European Commission)
The second lesson is consent. If someone’s face, voice, or persona is being cloned, there should be a clear rule: no consent, no clone. That idea fits naturally with European data-protection logic and with U.S. concerns around impersonation, fraud, and rights of publicity, even though the legal systems differ. (European Commission)
The third lesson is child protection. China’s draft does not treat youth risk as a side issue. It specifically addresses addictive relational design and manipulative emotional bonding. Western debates often discuss “AI safety” at a high level, but this draft asks a more concrete question: what happens when a synthetic persona becomes a companion, confidant, or emotional authority for a young user? (European Commission)
The fourth lesson is that interactive AI may need to be regulated like a behavioral product, not only like content. That means attention to dark patterns, compulsive use, emotional manipulation, and engineered dependency. In that sense, the draft looks less like a simple media rule and more like a hybrid of platform regulation, child-safety law, and consumer protection. (European Commission)
Where the West should differ
China’s framework is tied to censorship, state oversight, and broad national-security language. Western democracies should not copy that part. The better lesson is narrower and more practical: regulate synthetic personas where they affect trust, consent, manipulation, and children. Keep the protections. Drop the ideological control. (European Commission)
So the real question is not whether the West should become “more like China.” It should not. The real question is whether democracies can build equally clear rules for digital humans while protecting free expression, due process, and accountability.
That challenge is now on the table.
Questions for classroom discussion
- Is a visible “digital human” label enough, or will people still form emotional trust even when they know the figure is artificial?
- Should cloning a person’s face or voice require explicit consent in all cases, including for parody, education, or art?
- Is there a difference between an AI chatbot and a “digital human” with a face, voice, and personality? If yes, why does that difference matter?
- Should minors be protected from AI companions that simulate friendship, intimacy, or family roles?
- When does helpful interaction become manipulation? What signs should we look for?
- Should the law regulate only false content, or also the design of emotionally persuasive AI characters?
- Can Western democracies create strong guardrails for digital humans without drifting into censorship?
- In education, should students always be told when a teacher, tutor, or classroom guide is partly synthetic?
Field report prompts for students
Ask students to choose one or two of these and write a short field report:
Prompt 1:
A digital human says it is “just here to help.” What kinds of power can still exist in that interaction?
Prompt 2:
Compare a human influencer and an AI influencer. Which one is easier to trust, and which one is easier to control?
Prompt 3:
Should a digital companion for teenagers be legal at all? Explain your answer with reasons.
Prompt 4:
Write about a future classroom where some guides, tutors, or presenters are digital humans. What could improve, and what could go wrong?
Prompt 5:
Is labeling enough to protect users, or do we also need limits on design, emotion, and persuasion?
Prompt 6:
Who owns a face, a voice, or a personality in the age of AI cloning?
