China Tightens Rules for AI Companions

The main update is clearer evidence of implementation effects: major platforms reportedly rolled back or shut down companion features as the rules took effect. The framing also tightens around disclosure, minors, and product designs built on persistent memory and continuous emotional engagement.

The story has shifted from a broad description of China’s emerging companion-AI rules to confirmation that the measures are now in force and are being actively enforced. The current version also adds a broader comparative frame, contrasting China’s approval-and-filing approach with U.S. and EU AI oversight models.

• Rules now explicitly in force since July 15, 2026.
• Safety reviews required after major changes and user-threshold triggers.
• Sensitive companion-interaction data banned from training without consent.
• Shanghai removed more than 14,000 non-compliant agents.
• U.S. and EU AI oversight models added for comparison.

China’s AI story has shifted from a broad move toward unified regulation to an immediate enforcement phase for anthropomorphic companion services. The new rules create concrete operating constraints and have already triggered reported product rollbacks, while the planned unified AI law remains the longer-term consolidation effort.

• Anthropomorphic AI rules took effect on July 15, 2026.
• The rules target sustained emotional interaction services, not ordinary assistants.
• Users under 18 are barred from virtual intimate relationship services.
• App stores must verify filings and can suspend or remove noncompliant services.
• ByteDance, Alibaba, Tencent, and NetEase reportedly rolled back companion features.

The story now places greater emphasis on China moving toward a first comprehensive AI law, rather than just a broader national framework. The policy-analysis angle is also slightly reframed as evidence of a still-uneven governance system that Beijing is tightening under state-control and U.S.-China competition pressures.

The material centers on how China is structuring AI governance through central-level policy design, quantitative policy evaluation, and movement toward a unified AI law. Across the texts, the focus is less on individual AI applications than on the coherence, coverage, and institutional design of the regulatory system. The dominant finding is that China’s framework is becoming more systematized, while gaps remain in long-term planning, coordination, and policy instrument balance.