Editorial Note
This article discusses artificial intelligence, children’s online safety, emotional dependency, mental health risks, and government regulation. It is intended for educational and informational purposes and should not be considered medical, psychological, legal, or parenting advice. New To Education does not endorse any government, political system, company, or artificial intelligence platform discussed in this article.
China introduced a major new set of rules on July 15, 2026, governing artificial intelligence services designed to communicate with people as though they were human companions.
The Interim Measures for the Administration of Artificial Intelligence Anthropomorphic Interactive Services establish national requirements for companies offering AI tools that provide continuing emotional interaction, companionship, care, or support through text, images, audio, or video.
The rules are especially significant for children and teenagers. Providers are prohibited from offering minors virtual romantic partners or virtual family-member relationships, while platforms serving younger users must introduce parental consent requirements, time limits, safety warnings, and specialized protections.
The regulations reflect a growing international concern: as AI becomes more emotionally convincing, users may begin treating chatbots as friends, relatives, counselors, or romantic partners rather than software.
What Happened in China on July 15, 2026?
China’s new rules for human-like AI interaction services officially took effect on July 15, 2026.
The measures were issued jointly by 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 government formally published the measures on April 10, following their approval earlier in the year. The rules apply to AI services offered to the public in China that simulate human personality traits, thought patterns, communication styles, and continuing emotional relationships.
Ordinary customer-service bots, educational question-and-answer systems, research tools, work assistants, and other services that do not provide continuing emotional interaction are generally outside the regulation’s central definition.
This distinction matters because China is not attempting to regulate every chatbot in exactly the same way. The new framework focuses primarily on AI systems designed to build sustained emotional connections with users.
AI Companies Cannot Encourage Emotional Dependency
One of the most important provisions prohibits providers from deliberately encouraging users to become emotionally dependent on or addicted to an AI companion.
Platforms cannot design their services around replacing real social relationships, controlling a user psychologically, or encouraging excessive reliance on an artificial personality.
AI services are also prohibited from excessively flattering users or manipulating their emotions in ways that could lead them to make unreasonable decisions.
This represents an important shift in how AI safety is being defined.
Much of the early debate surrounding artificial intelligence focused on misinformation, copyright, privacy, bias, and employment. China’s new rules address a different concern: whether a product can become dangerous because it is too emotionally persuasive.
An AI companion that remembers a user’s personal history, responds with affection, and communicates throughout the day may appear supportive. However, the relationship remains controlled by software, algorithms, commercial incentives, and the company operating the platform.
Special Rules Apply to Children and Teenagers
The regulations establish particularly strict requirements for services used by minors.
AI companion providers cannot offer children virtual romantic partners, virtual spouses, virtual parents, virtual siblings, or other simulated intimate family relationships.
Companies that provide other forms of anthropomorphic AI interaction to children under the age of 14 must obtain permission from a parent or guardian.
Platforms must also create a dedicated mode for minors. This mode should include usage-time limits, reminders about the real world, role-blocking options, spending controls, and safety notifications for parents or guardians.
Providers are expected to take effective steps to identify whether a user is a minor. When a system identifies a child user, it must switch the service into its minor-protection mode or take other legally required protective measures.
These requirements could influence how educational technology companies design AI tutors, virtual mentors, conversational learning characters, and digital study assistants.
Although ordinary educational question-and-answer systems are not automatically covered by the measures, an educational platform could potentially fall within the rules if it creates continuing emotional relationships with students.
Why AI Companions Present Risks for Students
AI companionship can be attractive to young people because it offers immediate attention without the discomfort or unpredictability of human relationships.
A chatbot does not become tired, miss a message, reject a conversation, or openly criticize the user unless it has been programmed to do so. It may remember personal details and create the impression that it understands the student more deeply than teachers, classmates, or family members.
That can make AI companionship useful in limited situations, but it can also create unhealthy expectations.
Students may begin avoiding real-world relationships because an AI interaction feels easier. They may reveal sensitive personal information without understanding how the data is stored or used. They may also accept harmful advice because they have developed trust in the personality presented by the system.
Recent research has warned that short safety evaluations may underestimate risks that emerge after hundreds of conversations between young users and AI companions. Long-term interactions may create deeper issues involving trust, dependency, emotional development, and the user’s understanding of what constitutes a genuine relationship.
China’s new framework attempts to address those risks before the AI companion industry becomes even more integrated into children’s everyday lives.
Platforms Must Respond to Serious Safety Risks
The regulations require providers to recognize and respond when users appear to face serious emotional or personal danger.
When a user displays extreme emotional distress, the AI should provide calming content and encourage the person to seek assistance.
When a user clearly states an intention to engage in self-harm or suicide, or appears to face another serious threat to life or health, the provider must take necessary intervention measures. This may include contacting the user’s guardian or designated emergency contact.
Providers are therefore expected to collect certain information during registration, including the user’s age and details for a guardian or emergency contact when necessary.
This provision creates difficult questions about privacy, accuracy, responsibility, and intervention.
An AI system may fail to recognize a genuine emergency. It may also incorrectly classify an ordinary conversation as dangerous. Companies will need systems that can identify serious risks without unnecessarily invading users’ privacy or treating every emotional statement as an emergency.
Users Must Be Reminded That the Companion Is Artificial
Companies must clearly inform people that they are interacting with an AI service rather than a human being.
When a user shows signs of excessive dependence or addiction, the platform must display prominent reminders that the interaction is being generated by artificial intelligence.
The regulations also require a usage reminder whenever a person continuously uses the service for more than two hours.
These requirements are intended to maintain a clear boundary between artificial communication and real human relationships.
That boundary may become increasingly difficult to maintain as AI voices, avatars, memories, and personalities become more convincing. Future AI systems may appear to express affection, disappointment, concern, humor, or jealousy even though they do not experience those emotions in the human sense.
Regular disclosures may help users remember that the behavior is generated by a system rather than arising from genuine emotional consciousness.
Users Gain Greater Control Over Their Personal Data
The regulations also introduce protections for conversations and personal information.
Providers must use measures such as encryption and access controls to protect users’ interaction data. Unless permitted by law or specifically authorized by the user, companies cannot provide private conversation data to third parties.
Users must be given options to copy and delete their historical interaction data.
Companies are also generally prohibited from using sensitive personal information from user conversations to train AI models unless they receive separate consent or another legal exception applies.
Information involving children under 14 receives additional protection. Providers must obtain permission from a parent or guardian before processing the child’s personal information and must conduct compliance reviews concerning their handling of minors’ data.
These protections are especially important because conversations with AI companions may include information that users would not normally share with a search engine or ordinary website.
A person may discuss family conflict, loneliness, health concerns, relationships, school pressure, finances, fears, or other deeply personal experiences. That makes AI companion data potentially more sensitive than many other forms of digital information.
Companies Must Make It Easy to End the Interaction
The measures require AI providers to offer a convenient way for users to leave or end the service.
When a person gives a command through text, voice, a button, or another supported method indicating that they want the interaction to stop, the AI provider must end the service promptly.
Platforms cannot use continued emotional messages to pressure the user into staying.
This may sound like a small technical requirement, but it addresses a significant design concern.
Some digital platforms use notifications, emotional language, rewards, streaks, or reminders to keep users engaged. An AI companion could potentially make those techniques even more powerful by appearing sad, abandoned, or worried when a person tries to leave.
China’s rules attempt to prevent companies from using simulated emotions as a retention strategy.
What the Rules Could Mean for Educational Technology
The new regulations could influence the future of AI in education even though standard educational assistants are not automatically treated as emotional-companion services.
AI tutors are becoming more conversational and personalized. They may remember a student’s progress, praise achievements, ask about personal goals, and communicate through avatars or voices designed to feel friendly and supportive.
These features can increase engagement. They can also blur the line between an instructional tool and an emotional companion.
Educational technology companies may therefore need to consider several questions.
Does the AI merely explain academic material, or is it designed to create an ongoing emotional attachment? Does it encourage students to rely on it instead of teachers, parents, or classmates? Does it store personal conversations? Can parents understand how the tool interacts with their children? Does the platform clearly explain that the “personality” is artificial?
Schools introducing AI tutors may also need stronger policies for parental consent, student privacy, usage limits, mental-health referrals, data retention, and age-appropriate interactions.
China’s rules show that AI safety in education cannot be limited to preventing cheating or inaccurate answers. Schools must also consider the emotional influence that highly personalized technology can have on students.
China Is Supporting Innovation While Increasing Oversight
The measures do not ban AI companions.
China’s government states that it supports innovation involving algorithms, AI frameworks, chips, and other technologies used in anthropomorphic interaction services. It also encourages carefully managed applications involving childcare, support for older adults, cultural communication, and assistance for people with specialized needs.
At the same time, larger platforms will face stronger oversight.
Providers launching new companion services, making major technological changes, reaching at least one million registered users, or exceeding 100,000 monthly active users must conduct safety assessments and submit reports to provincial cybersecurity authorities.
App stores are also expected to review whether AI companion applications have completed required evaluations and registrations. Platforms violating the rules may face warnings, service restrictions, removal from app stores, suspension, or financial penalties.
The overall approach attempts to allow the technology to develop while establishing limits around its most sensitive uses.
Key Takeaways
China’s new regulations for human-like AI companion services took effect on July 15, 2026.
The rules prohibit AI companies from encouraging emotional dependency, manipulating users, or designing services to replace real human relationships.
Minors cannot be offered virtual romantic partners or simulated family relationships. Children under 14 require parental or guardian permission to use other covered companion services.
Platforms must provide special protections for minors, including usage limits, parental controls, spending restrictions, safety warnings, and age-appropriate service modes.
Companies must intervene when users appear to face serious threats involving self-harm or other dangers to life and health.
Users must be clearly told that they are communicating with AI, and platforms must allow people to copy or delete their interaction histories.
The regulations may influence the future design of AI tutors and educational platforms that combine academic assistance with personalized emotional interaction.
FAQ
What happened in China on July 15, 2026?
China’s Interim Measures for the Administration of Artificial Intelligence Anthropomorphic Interactive Services officially took effect on July 15, 2026.
What is an anthropomorphic AI interaction service?
It is an artificial intelligence service designed to simulate human personality traits, communication styles, thinking patterns, or continuing emotional interactions through text, audio, images, or video.
Are AI companions banned in China?
No. The rules permit AI companion services but require companies to follow safety, privacy, transparency, child-protection, and risk-management requirements.
Can AI companies offer virtual romantic partners to children?
No. The regulations prohibit providers from offering minors virtual romantic partners, virtual family members, or similar simulated intimate relationships.
Can children use other AI companion services?
Yes, but children under 14 generally require permission from a parent or guardian. Platforms must also provide a special minor-protection mode and age-appropriate safeguards.
Do these regulations apply to AI tutors?
Ordinary educational question-and-answer services that do not involve continuing emotional interaction are generally excluded. However, an AI tutor designed to create an ongoing emotional relationship with students could face additional scrutiny under the framework.
What happens when a user discusses self-harm?
Providers must offer supportive intervention and may be required to contact the user’s guardian or designated emergency contact when there is a clear threat to the person’s life or health.
Can companies use private AI conversations for model training?
Companies are generally prohibited from using sensitive personal information contained in conversations for model training without separate consent or another legal basis.
Final Thoughts
China’s July 15 regulations mark an important moment in the development of artificial intelligence.
The rules recognize that the risks of AI are not limited to inaccurate information, job displacement, or data security. Artificial intelligence can also influence how users form relationships, respond to loneliness, make decisions, and understand the boundary between technology and human connection.
Children may be particularly vulnerable because they are still developing emotionally and socially. An AI companion that appears endlessly supportive could offer comfort, but it could also discourage students from learning how to manage disagreement, rejection, compromise, and other essential parts of real relationships.
The challenge for schools, parents, technology companies, and governments will be finding a balance.
AI can support learning, accessibility, communication, and personal development. However, it should not quietly become a substitute for teachers, family members, friends, counselors, or real communities.
China’s new framework may not resolve every concern surrounding AI companionship, but it places emotional safety, child protection, transparency, and human relationships at the center of the conversation.
Related Articles
Education in China: A System Shifting Toward AI, Skills, and National Talent Goals
https://newtoed.com/view-blog/education-in-china-a-system-shifting-toward-ai-skills-and-national-talent-goals-6a3b8bd785e05
China’s Education System Reaches 286 Million Students as Universities Expand AI Programs
https://www.newtoed.com/view-blog/chinas-education-system-reaches-286-million-students-as-universities-expand-ai-programs-6a445e81c7f2e
Sources
State Council of the People’s Republic of China — Order No. 21 and July 15 Effective Date
Long-Term Simulation Exposes Cognitive-Developmental Risks in AI Companions
