Technology & Governance

China's New Rules for AI Companions: Guarding Privacy and Preventing Emotional Addiction

As companion robots and emotionally intelligent chatbots go mainstream, China has introduced its first dedicated rules for AI anthropomorphic interaction services, targeting privacy abuse, emotional manipulation, and tech ethics.

• The new measures, effective July 15, 2026, were jointly issued by China's cyberspace regulator and four other agencies, covering privacy protection, preventing emotional dependence, and tech ethics review.
• Companies are barred from using anthropomorphic design to blur human-machine boundaries, induce addiction, or exploit users' vulnerability to manipulate them into harmful decisions.
• Higher protection standards apply to minors and the elderly, and companies must ensure responsibilities are traceable rather than blaming "autonomous machine behavior."

Sometime in mid-2026, a full-size bionic companion robot from a major Chinese brand went on sale — and demand far exceeded expectations. Before the launch event had even ended, the company announced that pre-orders across all channels had surpassed 13,000 units. The product's popularity was a vivid demonstration of a growing market: emotionally intelligent machines designed to keep people company.

But popularity also brought scrutiny. As these "mechanical companions" multiply, so do questions about data privacy, emotional manipulation, and where the line between human and machine should be drawn. China has responded with a groundbreaking piece of regulation: the Interim Measures for the Management of AI Anthropomorphic Interaction Services, which took effect on July 15, 2026.

Issued jointly by the Cyberspace Administration of China, the National Development and Reform Commission, the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Administration for Market Regulation, the measures draw clear red lines for AI products and services that simulate human-like interaction. They cover three core areas: user privacy and personal information protection, prevention of emotional dependence and addiction, and scientific and ethical review of technology.

Legal experts describe a set of ethical bottom lines that the rules aim to enforce. The human-machine boundary must not be deliberately blurred: providers cannot design products to replace genuine social interaction, control users' psychology, or induce addiction. Privacy and data protections require companies to collect personal information only within appropriate scope, purpose, and necessity. Emotional manipulation is prohibited — firms cannot exploit users' weaknesses, such as illness or age, to push them into unreasonable consumption or decisions that harm their own interests.

Special emphasis is placed on vulnerable groups. Minors and the elderly are entitled to higher protection standards. The rules also stress responsibility traceability: in cases of infringement, product defects, data leaks, or algorithmic errors, responsibility must be assigned to the actual producers, sellers, and operators — companies cannot wave away liability by citing "autonomous machine behavior."

Experts recommend that governance be built into the design stage, not retrofitted later. For home-use companion robots, they advocate "privacy by default": sensors should not capture data unless necessary, processing should happen locally whenever possible, and camera and microphone use should have clear, easy-to-use on-and-off controls. Companies should obtain explicit separate consent before using personal data for purposes like model training, rather than burying such uses in blanket terms and conditions.

The regulation is part of a broader effort to balance innovation with safety. As humanoid robots and emotionally intelligent software become more common, the challenge is to protect users — especially the young and the elderly — without stifling a promising industry. China's approach, still early and evolving, offers a notable template: set clear ethical and safety guardrails first, then let the technology grow within them.