China has started fining artificial-intelligence platforms under its first nationwide rule for AI companion apps. Penalties reach up to CNY 200,000 for violations tied to user harm. The China AI companion fines follow rules that took effect July 15. Five agencies issued those rules together, calling them the Interim Measures for the Administration of AI Anthropomorphic Interactive Services. Within days, two of China’s largest AI platforms shut down their customizable AI-persona features entirely.
What China’s new AI companion rule actually bans
The regulation targets services built around sustained emotional interaction. These services simulate human personality traits, thinking patterns and communication styles. The rule excludes customer-service bots, knowledge Q&A tools, workplace assistants, and education or research tools. Those tools qualify only if they avoid sustained emotional engagement with users.
How the China AI companion fines are being enforced
The Cyberspace Administration of China oversees compliance and enforcement. Simple violations draw fines up to CNY 100,000. That figure rises to CNY 200,000 when a violation harms a citizen’s life, health or safety. Authorities can also order a service suspended for refusal to rectify, or in serious cases.
Why two of China’s biggest platforms shut down personas overnight
Two of China’s largest AI platforms shut down their customized AI-persona services within about ten days of receiving a cease-operation notice, according to reporting on the rollout. Both companies moved quickly once enforcement of the rule began on July 15.
What counts as emotional AI under the rule
Five agencies co-issued the Interim Measures on April 10, 2026. They are 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 defining line is “sustained emotional interaction.” That phrase covers services built around ongoing simulated relationships, not simple task completion.
How this compares with AI rules elsewhere
China’s approach to emotionally interactive AI stands apart from the European Union’s AI Act. That law focuses more broadly on risk classification and transparency for AI systems in general. It also requires that users learn when they’re interacting with AI at all. China’s rule instead singles out one specific category of service: sustained emotional simulation. It builds a dedicated framework around that category, backed by fines and the power to force a shutdown. Analysts tracking global AI regulation see the two approaches as complementary case studies. Both reflect how governments are choosing to regulate the same underlying technology. The underlying concern is shared across jurisdictions: emotionally realistic AI can shape vulnerable users’ behavior.
What happens as enforcement continues
The Cyberspace Administration has signaled ongoing monitoring, not a one-time compliance sweep. More fines are likely as enforcement continues. Regulators and analysts elsewhere are watching China’s framework closely. Many see it as one of the first national approaches of its kind to emotionally interactive AI.
What users of these apps can expect next
Users who relied on the shut-down persona features have had to switch to alternative apps. Others have accepted more limited functionality from their existing ones. Platforms that continue offering AI companion features in China now face a clear compliance bar. They must avoid sustained emotional simulation, or operate strictly under the new rule’s conditions. Some smaller developers may choose to exit the emotionally interactive AI category entirely. Building the compliance infrastructure the rule requires costs money that not every developer wants to spend. That calculation could reshape which companies stay active in this corner of China’s AI market over the coming months.
Why enforcement moved so quickly
Chinese regulators have shown a pattern this year of moving fast once a new AI rule takes effect. They tend not to allow a long grace period. The cease-operation notices issued to major platforms within days of the July 15 deadline reflect that same approach. Companies operating in China’s AI sector now treat compliance deadlines from Beijing as firm from day one. Few treat them as an opening position for negotiation. That approach contrasts with the phased rollout the EU has used for its own AI Act, where high-risk obligations only apply years after the law’s initial transparency requirements took hold.
China’s AI companion rules: common questions
What is the Interim Measures for the Administration of AI Anthropomorphic Interactive Services?
China’s first nationwide rule governing AI services designed for sustained emotional interaction with users. Five government agencies co-issued it.
When did the rule take effect?
July 15, 2026.
Who enforces the rule?
The Cyberspace Administration of China, working with four partner agencies that co-issued the regulation.
What are the penalties for violating it?
Fines up to CNY 100,000 for simple violations. That rises to CNY 200,000 when a violation harms a citizen’s life, health or safety, with possible service suspension in serious cases.
Are customer-service chatbots covered by the rule?
No, as long as they avoid sustained emotional engagement. The rule specifically excludes customer service, Q&A, workplace and education tools that meet that condition.
Could other countries adopt similar AI companion rules?
It’s too early to say. But analysts note China’s framework is one of the first of its kind, and regulators elsewhere are watching how it plays out.