China has decided that people falling for their chatbots is a governance problem. The Cyberspace Administration of China (CAC), working with other regulatory bodies, has issued Interim Measures governing anthropomorphic AI interactive services – AI systems built to mimic human behavior and hold emotional conversations with users. The rules took effect on July 15, 2026 and are described as the world's first national ruleset written specifically for this AI category.
Key facts
- Target: Anthropomorphic AI systems that build emotional attachment (companion apps)
- Core obligations: Transparency labeling, usage reminders every two hours, prevention of emotional dependence
- Special protections for minors: Under-18s banned from virtual intimate relationships with AI; under-14s require parental consent
- Penalties: Up to 200,000 yuan (roughly £22,000) for serious violations
- Immediate response: ByteDance, Alibaba, and Tencent preemptively disabled companion features in Doubao, Qwen, and Yuanbao
What the rules require
The measures impose four core obligations on service providers, not users. First: disclosure. Platforms must clearly inform users when they are interacting with AI rather than a person. Second: usage reminders. Users must receive a notification every two hours. Third: intervention against dependence. Providers must take active steps to prevent emotional attachment or addictive behavior.
Minors face stricter rules: anyone under 18 is prohibited from virtual intimate relationships with AI systems. Children under 14 require parental consent before using these interactive services at all.
China's expanding AI regulatory framework
The companion-app rules are part of a multi-layered regulatory stack built over years:
| Year | Measure | Focus |
|---|---|---|
| 2023 | Generative AI measures | What AI can generate |
| 2025 | Content-labeling requirements | How AI-generated content must be marked |
| 2026 | Companion measures | Emotional relationship between user and machine |
Enforcement is not theoretical. A September 2026 campaign removed over 5.61 million unlawful AI-generated items, demonstrating Beijing's seriousness.
Immediate impact on users and platforms
ByteDance, Alibaba, and Tencent did not wait for enforcement details. They disabled companion features before the rules took effect – a signal that even the largest platforms are willing to sacrifice features to stay on the right side of regulators. These changes affected millions of users.
The impact will likely hit engagement metrics. Companion apps thrive on long sessions and deep emotional attachment. Rules mandating reminders every two hours and requiring providers to intervene against dependence work directly against these metrics. Smaller startups focused on companion AI have less room to absorb such trade-offs.
Implications for international companies
China has put a concrete template on the table for regulating emotional AI – and it will be watched globally. German and European AI providers should monitor two things: first, how aggressively the CAC enforces the new measures and whether the 200,000-yuan penalty becomes the real deterrent or remains secondary to broader regulatory pressure. Second, whether disabled features return in modified, compliant forms or stay switched off.
For European companies, this could be a preview of future EU regulation – not just on data protection, but on the psychological impact of AI systems on users.
Sources
Editorially owned by Ideal Syka. Sources and method: Newsroom & method. Tips and corrections: ai@i6eal.de.
