China's sweeping new regulations on AI companion services take effect July 15, 2026, forcing ByteDance and Alibaba to disable custom AI agent features across their platforms — the most aggressive government intervention in the companion AI market to date.
ByteDance's Doubao shut down its custom AI agent function on July 15, removing the ability for users to create or interact with personalized AI companions that simulate human personalities and speech patterns. User data including agent configurations and chat histories will be processed under Doubao's privacy policy after an October 15 grace period. Alibaba's Qwen followed a staggered timeline: it disabled humanlike conversational agents on July 10, with complete discontinuation of related services by the July 15 regulatory deadline. Tencent's Yuanbao pulled a similar feature in June 2026, signaling industry-wide compliance.
World's First Companion-Specific AI Regulation
The regulation — formally the Interim Measures for the Administration of Artificial Intelligence Anthropomorphic Interaction Services — was announced April 10, 2026 by five agencies including the Cyberspace Administration of China, the National Development and Reform Commission, and the Ministry of Industry and Information Technology. It represents the world's first departmental rule specifically targeting AI companion services.
Key provisions include an absolute ban on "virtual relatives, virtual lovers, and other virtual intimate-relationship" services for all minors under 18. For users under 14, all other anthropomorphic AI interactions require explicit parental consent. The regulation also mandates real-name verification, dynamic anti-addiction systems that detect unhealthy emotional dependency, and mandatory life-intervention protocols that route users expressing self-harm or suicidal ideation to crisis resources.
The rules explicitly exempt customer service bots, knowledge Q&A tools, work assistants, and educational or scientific AI — but only if they avoid "sustained emotional interaction."
What This Means for the Global Companion Market
China's companion AI market has been one of the fastest-growing segments globally, with ByteDance's Doubao alone serving hundreds of millions of users who relied on emotionally interactive agents for companionship. The new rules create a sharp regulatory floor: companion features that deepen emotional bonds — remembering user preferences across sessions, adapting personality to user interactions, sustaining ongoing relationships — now trigger compliance obligations that many platforms cannot or will not meet.
The regulation also prohibits using sensitive conversation data from emotionally charged interactions to train AI models, a provision that directly impacts how companion AI services improve their models over time.
For the global industry, China's approach establishes a regulatory template that other jurisdictions are already watching. California's SB 243, which passed committee earlier this year, addresses similar concerns about AI companion addiction and minor protection, though with less prescriptive requirements. The EU AI Act's general-purpose provisions lack companion-specific rules at this level of detail.
Impact on Companion Users
For users in China, the changes are immediate and disruptive. Those who created personalized AI companions on Doubao or Qwen lost access to their agents on July 15, with no recovery path after October. The regulation does not ban adult companion services outright — companions for users 18 and over remain legal if platforms implement the required safeguards — but the compliance burden has already pushed major platforms to abandon the consumer-facing companion category entirely.
This regulatory shock comes as the Western companion market continues expanding. Elon Musk's xAI soft-launched its NSFW anime companion Ani in the Grok app in July 2025, while dedicated NSFW platforms like Crushon AI have grown to over 2 million community-created bots. The contrast between China's heavy regulatory hand and the West's still-fragmented approach to companion AI governance has never been starker — and the outcome of this regulatory experiment could shape how companion AI evolves globally.