China has confirmed a deadline. July 15, 2026 is when the “Interim Measures for the Management of Anthropomorphic AI Interaction Services” take effect. That date is confirmed – not estimated.
The rules were jointly issued by four ministries: 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. This is the final rule. Geopolitechs’ regulatory analysis confirms the issuing bodies and effective date directly from the regulatory text.
One attribution note for context: China’s Cyberspace Administration issued the December 2025 draft of these rules. The final rule was issued by the four ministries named above. These are distinct actions from distinct bodies. Any framing that attributes the final rule to “China’s CAC” is incorrect.
The Scope Decision Is the News
The December 2025 draft used broad definitions that raised compliance questions across the AI industry, customer service bots, virtual assistants, and productivity tools were all potentially within scope. The final rules resolve that uncertainty.
Per the Geopolitechs analysis, the final rules narrow the regulatory scope to “sustained emotional interaction services.” The exclusions are explicit: customer service bots and productivity tools are out of scope. The rules target AI companions, social AI platforms, and any service where ongoing emotional engagement is the core product function.
That line, “sustained emotional interaction”, is the compliance boundary. Teams building AI companions or social AI features need to assess whether their product falls inside it. Teams building productivity tools with conversational interfaces can proceed with reasonable confidence that they’re outside it, absent a material change in product design.
Youth Protections
According to legal analysis by Mayer Brown, the rules include an explicit prohibition on providing minors with virtual family member or romantic partner AI services. This element of the rules wasn’t in the confirmed fetched content, so Mayer Brown’s characterization is the attribution here rather than the regulatory text directly.
The prohibition extends the minor protection logic seen in China’s earlier platform regulations. If your product has any feature that could be characterized as providing a minor with a simulated intimate relationship, romantic, parental, or otherwise, that feature requires immediate review before July 15.
The 80-Day Window
July 15 is roughly 89 days from this publication date. For companies already operating in China’s AI companion or social AI market, that’s not a comfortable runway. Identity verification systems, youth mode implementations, and service design changes all require engineering time on top of legal analysis time.
The practical sequence: legal assessment of whether your product falls within “sustained emotional interaction services” (first 30 days), product and engineering gap analysis against the specific requirements (days 30-60), implementation and testing (days 60-89). That sequence leaves almost no margin.
TJS Synthesis
China’s scope narrowing is a meaningful signal beyond the immediate compliance question. The final rules reflect a deliberate regulatory design: target the highest-risk application category specifically rather than impose broad requirements across all conversational AI. The “sustained emotional interaction” qualifier is precise. It will generate edge cases, products that sit near the boundary will require judgment calls. But it’s more workable than the draft’s expansive framing. Companies outside scope should document why they’re outside scope. Companies inside scope have 89 days. Both groups are better served by this clarity than by the ambiguity the December draft left open.