China is tightening governance of the growing AI 'digital humans' sector, with draft rules from the Cyberspace Administration targeting consent, labeling, and safeguards to curb potential harms.

Industry players like Super Brain, which creates AI avatars of the deceased for grieving families, back regulation as a means to balance growth with standards.

The proposed rules prohibit creating deepfake clones from personal data without consent, require clear labeling of digital human content, and aim to shield children, protect social stability, and guard national security.

Analysts emphasize the tension between rapid AI adoption and the need to guard against misinformation, scams, and potential emotional or psychological harms.

Experts note regulators are taking a ‘develop first, regulate later’ approach, seeking to regulate as the sector expands rather than stifle innovation.

Rules bar content that endangers national security or incites subversion and forbid minors from engaging in virtual intimate relationships or forming extreme emotional bonds with digital humans.

Ethics debates have intensified after a widely shared episode where an elderly woman interacted with a hyper-realistic avatar of her deceased son, highlighting risks and the need for safeguards.

Industry figures argue these reforms are necessary to balance growth with risk management, including preventing deepfake misuse and protecting consumers.

Beijing aims to accelerate AI adoption within a controlled framework that preserves sovereignty and political objectives, reflecting a 'develop first, regulate later' stance.

Penalties under CAC rules could range from 10,000 to 200,000 yuan, and public comments on the draft are open through early May.

The draft rules also prohibit digital humans from generating or distributing content that endangers national security or subverts state power.

Personal stories illustrate the emotional impact of digital human avatars on grieving individuals seeking comfort and motivation.