China Regulates Digital Humans With Draft Rules
China's Cyberspace Administration (CAC) published draft rules to govern 'digital humans' — virtual avatars and interactive AI personas — imposing mandatory labeling, consent requirements, and content limits. The draft requires prominent "digital human" labels, bans virtual intimate relationships for users under 18, prohibits using a real person's identifiable traits without consent or to circumvent identity checks, and restricts content that threatens national security, incites secession, or promotes discrimination. The rules are open for public comment until May 6 and form part of a broader Chinese push to align AI services with state priorities and safety standards. Providers operating in or for China must reassess consent, identity-verification, content-moderation, and design practices.
What happened
The Cyberspace Administration of China (CAC) issued draft regulations to govern ‘‘digital humans’’ — interactive AI-generated personas and avatars — and opened the rules for public comment through May 6, 2026. The measures mandate visible "digital human" labels, forbid creating digital humans using another person's identifiable traits without consent, and bar virtual intimate-relationship services targeting under-18s. The draft also prohibits using virtual humans to bypass identity-verification systems and restricts dissemination of content that endangers national security, promotes secession, or incites discrimination.
Technical context
Digital humans are a convergent stack: generative models (text, speech, image/video), persona-conditioning layers that simulate thinking and affect, and realtime orchestration for interactive sessions. Regulatory levers in the draft map to those layers: labeling impacts user experience and UI/output pipelines; consent and likeness restrictions affect training data provenance, dataset curation, and model fine-tuning; identity-verification bans require integration of robust authentication flows and server-side checks; content restrictions demand stronger safety classifiers, prompt-engineering guardrails, and human-in-the-loop moderation.
Key details from sources
- •The CAC requires prominent "digital human" labels on all virtual-human content and continuous display in presentation areas. (Reuters)
- •Providers cannot use others' personal information to create digital humans without consent, and must not allow virtual humans to bypass identity verification. (Reuters; Xinhua)
- •Services that create "virtual intimate relationships" for people under 18 or otherwise encourage addictive behavior among minors are prohibited. (Reuters)
- •Content that threatens national security, incites subversion, or promotes secession, ethnic discrimination, sexual suggestiveness, horror, or cruelty must be prevented; providers are encouraged to intervene where users show self-harm risks. (Reuters; Mayer Brown)
Why practitioners should care
These draft rules change product, data, and compliance requirements for teams building interactive AI personas for China-facing users. Expect immediate implications across three domains:
- •data governance — provenance, consent capture, and takedown workflows for likeness and training material
- •runtime safeguards — persistent labeling, identity-verification checks, and stricter safety classifiers for content categories flagged by regulators
- •product design and UX — age-gating, limitations on relationship-simulating features, and escalation paths for user distress
Legal and compliance teams must interpret how the rules interact with existing IP, privacy, and cybersecurity obligations; external counsel will be needed to map obligations into operational controls.
What to watch
- •Final text and enforcement mechanisms after the May 6 comment deadline: whether penalties, audits, or licensing regimes are specified.
- •Technical standards for labeling, identity verification, and accepted methods to demonstrate consent for likeness use.
- •How broadly the definitions of "digital human," "virtual intimate relationship," and "addictive services" are interpreted in enforcement and litigation.
Scoring Rationale
China's draft rules impose concrete operational and compliance requirements on teams building interactive AI personas for a major market. The rules materially affect data provenance, identity-verification, content-moderation, and product features; timely attention is needed. The story is recent but past the three-day freshness window, reducing urgency slightly.
Practice interview problems based on real data
1,500+ SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
