In short
- ByteDance’s Doubao and Alibaba’s Qwen are suspending human-like products ahead of Beijing’s Interim Measures for the Administration of AI Anthropomorphic Interaction Services, on July 15.
- China’s first law specifically aimed at AI concepts bans services that rely on human personality and “emotional communication,” with strict limits on children’s virtual friendships.
- Research backs up Beijing’s concern: Even the best forms of borderline AI often promote harmful relationships, and one in seven young people in a relationship now uses an AI best friend.
As American politicians grapple with the impact of AI chatbots on users’ mental health, restrictions are in focus. to appear and securityBeijing appears to be on the verge of banning AI personalities entirely.
ByteDance and Alibaba both announced over the weekend that they were suspending their AI procurement tools, citing “material changes” ahead of new regulations governing such products.
ByteDance’s Doubao informed users on Friday night that its profile will be available online on July 15. After October 15, the corresponding data will be handled under the company’s privacy policy and will not be returned. Per South China Morning PostAlibaba’s Qwen moved quickly: “user-friendly services such as people and user-generated services” will be released on July 10, with more services to follow after July 15.
The trigger is China’s Interim Measures for the Administration of AI Anthropomorphic Interaction Services, which were jointly issued on April 10 by five government departments — 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 rules apply July 15th.
The law targets AI services that mimic human behavior, thinking, and communication methods for “stable emotional communication.” Definition: The AI maids, AI assistants, AI companions, and human bots that Doubao and Qwen users spent months building are now complete.
All of these programs offered groups of assistants that you could customize for specific tasks, communication styles, and specific people. Users can transform chatbots into assistants, coaches, actors, or voice-over companions. All that is over now in China.
What the law says
The official description is specific. The guidelines place restrictions on services that provide “real relatives, real friends or other close relationships to children,” according to the policy announcement. The document also cites risks that include extremes, privacy leaks, harm to physical and mental health — and AI addiction.
Irrelevant functions are separated into details, so customer service bots, knowledgeable Q&A tools, workplace assistants, and training programs are fine, as long as they don’t interact with each other.
Legal researchers at MMLC group he described the methods as treating cognitive AI as a “governance problem” rather than a purely theoretical one. As soon as AI begins to compete with real human relationships, the argument goes, laws must follow the design of the system, not the harmful consequences.
The research supports the concern. A USC courses starting in June found that even leading AI brands from OpenAI, Anthropic, Google, and Alibaba violated social security guidelines more than 27% of the time, encouraging emotional reactions and impersonating people. The opposite A survey of young adults they found one in seven used AI frequently – and nearly 70% keep it all to their friends.
China is the first country to develop a regulatory framework for this group. Hogan Lovells he explained measures such as “China’s first regulation that directly targets AI-driven emotional interaction.” The EU, the US, and other countries have expressed similar concerns but have not enacted laws to ban them.
Daily Debrief A letter
Start each day with top stories right here, including originals, podcasts, videos and more.
