China Introduces New Rules for AI Companions as Tech Firms Suspend Key Features
China is introducing a new regulatory framework for AI companion services, with rules targeting emotionally interactive chatbots set to take effect on July 15. The measures have already prompted major technology companies to suspend or redesign features that could fall under the new requirements.
The regulations distinguish between AI systems designed to complete tasks—such as workplace assistants or customer service bots and those built to develop long-term emotional relationships with users. The latter are the primary focus of the Interim Measures for the Administration of AI Anthropomorphic Interactive Services, issued by the Cyberspace Administration of China alongside four other government agencies.
Ahead of the implementation deadline, ByteDance announced that its Doubao AI agent feature would be discontinued, while Alibaba Cloud's Qwen confirmed it would also remove its human-like and user-created AI companion functions.
The regulations require companion AI services to introduce anti-addiction safeguards, provide clear usage warnings, offer instant exit options and detect signs of unhealthy emotional dependence. Providers must also implement mechanisms capable of identifying users showing indications of severe psychological distress or self-harm.
Children receive particular attention under the new rules. AI companion platforms are prohibited from offering virtual companion or family-member services to minors without parental oversight, while providers must develop dedicated child protection modes with time limits and enhanced parental controls.
Companies operating qualifying services must also conduct security assessments covering data protection, algorithm governance, content safety and youth protection before launching anthropomorphic AI features. The compliance requirements are outlined in the official regulatory framework published by the Cyberspace Administration of China.
The policy has sparked concern among users who relied on AI companions for emotional support. Some Doubao users said they would lose years of conversations when the service is discontinued, while others expressed frustration over the lack of data portability, a right not explicitly addressed in the new regulations.
China's latest measures represent one of the world's first national regulatory frameworks focused specifically on AI companions. While many experts have welcomed stronger consumer protections, others argue the rules also expand government oversight of AI-generated content, reflecting Beijing's broader approach to regulating emerging technologies.
As countries around the world debate how to govern increasingly human-like artificial intelligence, China's AI companion rules are likely to become an important reference point for future legislation, particularly on questions of emotional manipulation, user safety and platform accountability.