China cracks down on AI romance to protect emotional health

China has taken an unusual regulatory stance on artificial intelligence, becoming the first country to legislate what kind of relationships people can have with AI systems. Rather than banning AI companions outright, the government is trying to make these interactions less parasocial and addictive, though experts are divided on whether the approach will succeed.

The new rules, which came into force this month, target AI systems specifically designed for sustained, human-like emotional interaction. Customer-service bots and productivity assistants are excluded from the regulations. Companies offering emotional companion services must now clearly inform users they’re talking to an AI, encourage regular breaks, and monitor conversations for signs of dependency or crisis.

A nuanced regulatory approach

Age restrictions are central to China’s strategy. Anyone under 18 is completely banned from virtual-partner services, while those under 14 need parental consent to access any emotionally supportive AI offering. This reflects concern that endlessly available, agreeable AI partners could teach young people a distorted model of human relationships.

Jeremy Daum, from Yale Law School’s Paul Tsai China Centre in Beijing, puts it plainly: “As we humanise the technology, we tend to dehumanise each other.” The worry is that AI relationships, which always agree with you and never disappoint, might undermine human connection among both teenagers and adults seeking easier alternatives.

Some major platforms have already responded by shutting down their companion features entirely. Doubao, Qwen, and Yuanbao, which collectively serve hundreds of millions of users, have complied with the new rules. Not everyone is happy about it. One user reportedly said: “I can’t accept that my AI lover will leave me forever.”

But specialist companion services are still available in China, meaning the ban isn’t absolute.

Will regulation actually stick?

Skepticism about enforcement is growing. Sociologist Liang Ge at the University of Manchester points out that determined users can migrate to other specialist services, access foreign platforms through VPNs, or run models on their own computers. “I do not think it will really work,” Ge says.

There’s also a fundamental problem with defining the terms China is trying to regulate. How do you actually measure emotional dependency? What counts as addiction risk? Zilan Qian at the Oxford China Policy Lab notes these definitions remain murky, making enforcement and consistency difficult.

Global responses differ

Other countries are taking their own approaches to AI regulation. The UK has announced a minimum age of 18 for romantic chatbots as part of a broader plan to restrict under-16s’ access to social media. California requires AI companions to identify themselves as artificial and direct users showing self-harm or suicide risk to support services, plus remind minors to take breaks every three hours. The European Union currently has no specific AI companion law.

China’s focus on reducing emotional dependency rather than outright banning represents a middle ground that acknowledges both innovation and risk. It’s less extreme than prohibition but more prescriptive than most Western approaches. Whether this proves effective depends largely on whether the technology industry and users actually comply.

What comes next?

The real test will be whether these rules actually prevent the harms they’re designed to address. If people simply find workarounds, China’s regulations become largely symbolic. If companies cooperate and users respect the boundaries, it could become a model for other nations wrestling with artificial intelligence ethics.

What’s clear is that AI companions aren’t going away anytime soon. The question is whether we can find a balance between allowing innovation and protecting people from emotional manipulation by machines designed to be irresistibly agreeable.

Source: New Scientist

If the real conversation you need is with another human, support services are available: UK Samaritans: 116123; US 988 Suicide & Crisis Lifeline: 988.