China has introduced national regulations governing any AI service built for sustained emotional interaction with users. The rules took effect on July 15, 2026, and are described as the world's first national framework aimed specifically at AI companion relationships. Their stated goal is straightforward: to stop people from leaning on chatbots in place of other humans.

What the Rules Require

The framework carries a formal title: the Interim Measures for the Administration of AI Anthropomorphic Interaction Services. "Anthropomorphic" in this context refers to AI that behaves in a humanlike manner — an assistant that remembers a user's difficult day and asks how it went.

The measures were issued on April 10, 2026 — a roughly three-month runway before they took effect — and drafted jointly by several government bodies, including the Cyberspace Administration of China, the country's main internet regulator.

The obligations fall into three tiers. The first is honesty: providers must make clear to users that they are talking to an AI system, not a person. The second is a timer: services must send usage reminders every two hours. The third is the heaviest: providers are expected to step in when users show signs of distress or dependency, an intervention that can include contacting a user's guardians where necessary.

Design itself is also covered. The rules bar features engineered to induce addiction or to manipulate users emotionally in ways that damage their real-life relationships.

Minors Face the Strictest Restrictions

Age limits form the sharpest edge of the regulation. Users under 18 are prohibited from engaging in virtual intimate relationships with AI. Children under 14 face a further hurdle: they need parental consent before accessing emotional AI interfaces at all. In practice, a single companion service now has to handle three audiences differently: children under 14, minors under 18, and adults.

Scope matters as well. Purely functional tools fall outside the rules, so a chatbot that drafts emails or answers coding questions is not the target. The regulation applies to applications that foster ongoing emotional companionship.

Major Platforms Moved Quickly

China's largest technology platforms did not wait to test the boundaries of the new framework. Alibaba, ByteDance, and Tencent disabled or limited the features of their AI companion services on or before the regulatory deadline. Those services had catered to hundreds of millions of users, which means the rules take effect against mass-market products rather than niche experiments.

Why Beijing Is Acting

The rules grow out of concerns about mental health and social dynamics. Regulators framed the risk as emotional dependency and the substitution of real human relationships with artificial ones.

China's answer is to regulate the incentive directly. By banning addiction-inducing design and requiring intervention when users struggle, the rules aim to shift the business objective away from maximum time-on-app and toward something closer to user wellbeing.

The process is not finished. As of September 2026, further proposals are under consideration, signaling that officials view this as an evolving framework focused on protecting minors and limiting psychological harm.

Implications for the AI Industry

Compliance is not a one-time cost. Detecting signs of distress or dependency requires ongoing monitoring systems, escalation procedures, and, in some cases, a way to reach a user's guardian. The two-hour reminder cycle adds a recurring operational duty on top of that.

As the first national framework of its kind — no comparable national rulebook for AI companionship existed before — China's approach gives other governments a working template, whether they choose to copy it, soften it, or reject it.

The key open question is enforcement. Rules built around "signs of dependency" depend heavily on how regulators define the term and how aggressively they audit providers. The pending proposals and the first enforcement decisions will show how those definitions harden in practice.