China’s New Rules for AI Companions: What Your Business Needs to Know Before July 2026

Table of Contents
Your AI Buddy Just Got a Rulebook
Imagine your AI companion suddenly has to tell you it’s not human, explain how it uses your data, and stop collecting your emotional secrets without permission. That’s exactly what China’s new Interim Measures for the Administration of Anthropomorphic Interactive Services Based on AI will require starting July 15, 2026. If your company offers virtual friends, therapists, or even customer service avatars to users in China, this regulation is your new best friend—or your worst nightmare.
Who’s Affected? (Spoiler: Probably You)
These rules apply to any provider offering “anthropomorphic interactive services” to users in China. That includes chatbots with human-like personalities, virtual influencers, AI therapists, and even game NPCs that learn from your conversations. If your AI acts human, talks human, or makes users feel like they’re talking to a human, you’re in scope. No exceptions for foreign companies—if you serve Chinese users, you comply.
The Three Pillars: Transparency, Safety, Data Protection
1. Transparency: No More “I’m Totally Human”
Your AI must clearly disclose it’s AI. No pretending to be a real person. Think of it like a restaurant menu that must list allergens—except here, the “allergen” is deception. Users must know they’re talking to a machine, especially when the AI is designed to build emotional bonds. This is a direct response to concerns about users falling in love with or becoming dependent on virtual companions.
2. Safety: Keep It Clean and Kind
Content moderation gets serious. Your AI can’t generate harmful, illegal, or unethical responses. It must refuse to discuss topics like self-harm, violence, or political subversion. And if a user seems distressed, the AI should gently steer them to human help—not encourage dangerous behavior. This is like having a bouncer for your chatbot’s brain.
3. Data Protection: Your Secrets Stay Secret
Virtual companions often collect intimate data—your fears, your relationship problems, your late-night confessions. Under the new rules, this data must be minimized, encrypted, and deleted when no longer needed. You can’t sell it or use it for unrelated purposes. And users have the right to request deletion of their conversation history. Basically, treat user data like a diary you found on the street: don’t read it, don’t share it, and return it promptly.
Compliance Timeline: 18 Months to Get Ready
The measures take effect July 15, 2026, but don’t wait until June 2026. Start auditing your AI systems now. Map your data flows, update your privacy policies, and train your models to be transparent. The Chinese government has promised strict enforcement, including fines and service suspensions for non-compliance. Think of it as a fire drill—you don’t want to be the one still looking for the exit when the alarm rings.
Practical Steps for Your Business
- Conduct a gap analysis between your current practices and the new requirements. Focus on disclosure mechanisms, content filters, and data retention policies.
- Update your terms of service and privacy notices to include clear AI disclosure language. Make it obvious, not buried in legalese.
- Implement robust content moderation using both automated filters and human review for edge cases. Test your AI with sensitive prompts to ensure it responds appropriately.
- Review your data minimization practices. Only collect what’s necessary for the service, and set automatic deletion schedules for conversation logs.
- Consult with local legal experts in China. The regulatory landscape is evolving, and local guidance is invaluable. Check the Cyberspace Administration of China for official updates.
The Bigger Picture: A Global Trend?
China isn’t alone. The EU’s AI Act and various US state laws are also targeting AI transparency and safety. But China’s focus on anthropomorphic services is unique—it recognizes that human-like AIs pose special risks. If your business operates globally, consider adopting these standards everywhere. It’s easier to have one high-compliance bar than to juggle multiple lower ones.
So, grab your compliance checklist and a cup of coffee (or tea, if you’re feeling Chinese). The clock is ticking, and your virtual companion’s future depends on it.

NakedPact Editorial Committee
Article created by the NakedPact editorial team. Our mission is to analyze, simplify, and expose unfair terms and hidden risks in everyday contracts to protect citizens and consumers.
Sources and Legal References

Do you own a website?
Want to communicate your data processing transparency to your users? Dynamically use our badge and showcase your platform's compliance.
Recommended Readings
🛡️ Protect your rights with one click
Don't risk signing abusive clauses. Install the free NakedPact extension for Chrome or Firefox and instantly analyze any contract on the web.
Don't trust, verify.
Now that you know the risks, don't sign blindly. Upload your contract to NakedPact and let AI find the hidden clauses for you. It's 100% free.
Analyze Your Contract Now

