China is moving to tighten its grip on the fast-evolving world of human-like artificial intelligence, proposing new rules that would force AI providers to be far more transparent about how their systems interact with users. The draft regulations, released by the country’s top internet watchdog, aim to curb overdependence on AI while reinforcing security, ethics, and ideological boundaries.
If adopted, the rules would place China among the most assertive governments globally in regulating AI systems designed to mimic human behavior, from conversational bots to virtual companions.
What are China’s proposed rules on human-like AI?
The draft regulations were issued by the Cyberspace Administration of China, the powerful body that oversees internet and digital policy. They are open for public consultation until January 25.
Mandatory disclosure that users are interacting with AI
One of the most striking provisions requires AI platforms to repeatedly remind users that they are interacting with artificial intelligence, not a human.
Under the proposal:
- Users must be informed they are dealing with AI at the time of login.
- The disclosure must be repeated at least every two hours during continued use.
- Additional alerts are required if the system detects signs of user overdependence.
This goes beyond simple labeling. The emphasis on repeated reminders reflects growing concern that human-like AI systems can blur psychological boundaries, encouraging emotional reliance or misplaced trust.
Why is China focused on overdependence and transparency?
Chinese regulators have become increasingly wary of technologies that shape behavior at scale. Human-like AI, particularly conversational systems, raises concerns about emotional manipulation, misinformation, and excessive attachment.
Preventing unhealthy reliance on AI
By explicitly referencing “overdependence,” the draft rules acknowledge a risk that is being discussed globally but rarely regulated. Authorities appear concerned that users could:
- Treat AI systems as emotional substitutes for real relationships.
- Rely on AI for decision-making beyond reasonable assistance.
- Be influenced in subtle ways over long periods of interaction.
Requiring systems to actively monitor and flag overdependence suggests a preventative approach, rather than reacting after harm occurs.
Ethical and security requirements for AI providers
The proposals do not stop at user notifications. They impose a broad set of obligations on companies developing human-like AI.
Security and ethics reviews are mandatory
According to the draft:
- Providers must establish robust internal security systems.
- Ethical review mechanisms must be in place before launching new features.
- Human-like AI functions must undergo a formal security assessment.
Companies must also submit reports to provincial cyberspace authorities if they introduce such features, adding an additional layer of oversight before products scale.
User scale thresholds trigger extra scrutiny
China is also linking regulation to reach. The larger the audience, the heavier the compliance burden.
Reporting requirements based on user numbers
Under the proposed framework:
- Services must submit reports if they reach 1 million registered users.
- Alternatively, reporting is required at 100,000 monthly active users.
- Both thresholds apply specifically to human-like AI capabilities.
This reflects a risk-based approach, where systems with greater influence face closer monitoring.
Ideology and content controls remain central
As with most technology regulation in China, ideology is embedded directly into the rules.
Alignment with “core socialist values”
The draft requires AI systems to:
- Uphold “core socialist values.”
- Avoid generating content that could undermine national security.
- Refrain from publishing material authorities consider harmful to social stability.
This makes clear that human-like AI is not just a technical issue in China, but a political one. AI systems are expected to align with state priorities in both function and output.
How this fits into China’s broader AI strategy
China has identified artificial intelligence as a strategic industry central to future economic growth and global competitiveness. The country is investing heavily in AI research, infrastructure, and talent.
Innovation paired with control
The proposed rules highlight a dual-track approach:
- Accelerate AI development to boost productivity and international standing.
- Maintain tight governance to manage social, political, and security risks.
Rather than slowing AI adoption, the regulations aim to shape it in ways authorities believe are safer and more predictable.
How do these rules compare globally?
Globally, governments are grappling with how to regulate AI that feels increasingly human.
- The European Union’s AI Act focuses on risk categories and transparency, but does not mandate repeated user reminders.
- The United States relies more on sector-specific guidelines and voluntary commitments.
- China’s approach is more prescriptive, blending consumer protection, security oversight, and ideological enforcement.
The emphasis on overdependence warnings appears relatively unique, signaling where Chinese regulators see the most immediate risk.
What happens next?
The draft rules are currently open for public feedback until January 25. While revisions are possible, similar consultations in the past have often led to formal adoption with only limited changes.
If finalized, companies operating AI services in China will need to:
- Redesign user interfaces to include recurring AI disclosures.
- Build systems capable of detecting overdependence.
- Prepare for regular reporting and audits as their user base grows.
For foreign tech firms, compliance could become a significant operational challenge.
Why this matters beyond China
China’s regulatory decisions often ripple outward. Its market size means global AI companies may adapt products worldwide to meet Chinese standards, even if those rules are stricter than elsewhere.
The focus on transparency and psychological impact could also influence debates in other countries, where lawmakers are still struggling to define the risks of human-like AI.
TL;DR
China has drafted new rules to regulate human-like artificial intelligence, requiring repeated disclosures that users are interacting with AI and warnings when overdependence is detected. The proposals also mandate security reviews, ethical oversight, and ideological compliance, with stricter reporting requirements as user numbers grow. The move underscores China’s effort to balance rapid AI development with tight social and political control.