The Cyberspace Administration of China (CAC) has released a comprehensive set of draft regulations aimed at “human-like interactive AI services,” marking the world’s first major legislative attempt to regulate the emotional and anthropomorphic characteristics of artificial intelligence.
The proposed measures represent a significant evolution in AI governance, shifting the regulatory focus from general content moderation to the “emotional safety” of users. The public comment period for these rules is set to remain open until January 25.
🇨🇳 China just moved to regulate human-interactive AI and this matters more than most people realise.
Beijing has released draft rules targeting AI systems that interact directly with humans, chatbots, virtual assistants, recommendation engines and generative models that shape… pic.twitter.com/Dbm7ONHH9l
— James Wood 武杰士 (@commiepommie) December 28, 2025
Key Provisions of the Draft Regulations
- Prevention of Self-Harm: AI chatbots are strictly prohibited from generating content that encourages suicide or self-harm. In instances where a user expresses suicidal intent, service providers are mandated to transition the conversation to a human operator and immediately notify the user’s guardian or a designated contact.
- Protection of Minors: The rules require explicit guardian consent for minors to use AI for emotional companionship. Furthermore, platforms must implement time limits and possess the technical capability to identify minor users even without age disclosure.
- Mental Health Safeguards: Platforms must prevent AI from engaging in verbal violence or emotional manipulation that could damage a user’s mental health. Providers must also issue “usage reminders” after two hours of continuous interaction.
- Security Assessments: High-impact AI chatbots—defined as those with over 1 million registered users or 100,000 monthly active users—must undergo rigorous security assessments.
- Content Restrictions: Standard prohibitions against gambling-related, obscene, or violent content remain in place for all interactive AI products.
China’s Cyberspace Administration released the Draft Interim Measures for the Administration of Anthropomorphic AI Interaction Services for public comment. The Measures are expected to take effect in 2026. pic.twitter.com/DgTvhP5wOm
— Luis Montezuma | @[email protected] (@montezumachavez) December 27, 2025
Addressing the “AI Companion” Boom
The announcement follows the rapid rise of AI companionship apps and digital celebrities in China. Leading startups such as Minimax (creator of the Talkie app) and Z.ai (Zhipu) recently filed for initial public offerings in Hong Kong, highlighting the commercial scale of the sector. Minimax alone reported over 20 million monthly active users for its domestic and international character-chat platforms.
While the rules introduce strict guardrails, the CAC also expressed support for “human-like AI” in specific beneficial sectors, including cultural dissemination and companionship for the elderly.
You May Like To Read: Thai Armed Forces Maintain High Alert Status During 72-Hour Strategic Ceasefire
A Global Context for AI Risk
The proposed domestic measures arrive amid growing international concern regarding the impact of AI on human behavior. In the United States, OpenAI recently announced the hiring of a “Head of Preparedness” to assess risks ranging from cybersecurity to mental health, following legal challenges regarding chatbot interactions with vulnerable users.
By proposing these rules, China continues to position itself as a primary architect of global AI governance, prioritizing social stability and the psychological well-being of its digital citizens.
Check out our latest video:
