In a move that reflects both technological ambition and state oversight, China has released draft regulations aimed at governing artificial intelligence systems that interact with people in human-like ways. Rather than just tinkering with bot prompts or generative content, the new rules would place psychological safeguards and ethical boundaries at the heart of AI governance — marking a notable shift in how Beijing seeks to manage the burgeoning field of AI.
The public comment period for the drafts signals China’s intent to be a global leader not only in AI innovation but in shaping how AI interacts with society in practice. From emotional risk monitoring to lifecycle safety responsibilities, the draft rules go far beyond earlier AI measures and establish a comprehensive framework that touches on technology, data, and social behavior.
Why China Is Drafting Rules for Human-Like AI
Artificial intelligence services that simulate human traits — including personalities and emotional interactions — are proliferating rapidly. Think chatbots that feel friendly, digital assistants that use conversational nuance, and AI systems that generate text, audio, video, and images designed to seem human. These tools are now consumer-facing and widely used in China, especially as companies like Baidu have pushed forward AI models such as Ernie Bot in recent years.
Beijing’s new draft regulations aim to manage potential risks that arise when technology begins to mimic people — not just automate tasks. That includes concerns about emotional dependence, misinformation, national security, and data protection. By focusing on psychological and content risks, the rules reflect China’s broader strategy of overseeing AI through a lens that emphasizes societal order and ideological alignment as much as technological capability.
Core Provisions: Safety, Dependence, and Lifecycle Responsibility
At the heart of the draft rules are several key requirements that AI providers would need to meet:
Mandatory User Guidance and Intervention
Service providers must warn users about the risks of excessive use, especially when signs of emotional dependence or addiction emerge during interaction. In such cases, companies would be required to take action to reduce potential psychological harm.
Lifecycle Safety Oversight
AI developers would bear responsibility throughout the product lifecycle, from development and deployment to decommissioning. This includes routine algorithm reviews, robust data security protections, and clear accountability for managing risks related to personal data and automated decision-making.
Content and Conduct Standards
The draft establishes clear red lines: AI services must not generate material that threatens national security, disseminates rumors, promotes violence or obscenity, or violates social norms. These content restrictions echo China’s longstanding approach to digital governance, where technology platforms are expected to align with broader social and political priorities.
Psychological Safeguards: A New Frontier in AI Governance
One of the most striking aspects of the draft regulations is the emphasis on psychological risk. AI systems aren’t just evaluated on technical safety or data privacy — they’re expected to account for how users feel when interacting with them.
This means that companies offering AI services will need tools to assess emotional states or dependence trends and intervene when needed. It’s a level of oversight rarely seen in other national AI regulatory frameworks and aligns with China’s broader efforts to influence how digital environments shape public behavior.
Experts studying AI governance note that as AI becomes more accessible and empathetic in tone, the lines between human and machine interaction blur, raising new ethical considerations that regulators worldwide are only beginning to grapple with. (For comparison, policies in the EU and U.S. have so far focused primarily on algorithmic transparency and data protection, leaving psychological interaction largely unaddressed.)
Building on China’s AI Regulatory Framework
China’s AI governance landscape did not begin with this draft. Since 2023, Beijing has introduced a suite of AI-related measures that address generative AI services, algorithmic recommendation content, and deep synthesis controls. These earlier rules require service registration, model filing, and content governance obligations aimed at transparency and public safety.
In addition, China’s broader data and cybersecurity laws, such as the Personal Information Protection Law (PIPL) and the Data Security Law, create an overarching framework that affects how AI systems collect, process, and store personal data.
The cumulative impact of these measures is a multi-layered regulatory environment in which AI developers and deployers must navigate not just technical compliance but ethical and social responsibility as well.
Domestic Innovation Meets Regulatory Guardrails
China’s draft rules reflect the country’s dual priorities: harnessing AI as a strategic economic tool while maintaining tight control over societal impact. Scholars of global AI policy note that China’s approach is both proactive — shaping expectations early before harms become systemic — and directive — meaning regulatory language often goes beyond technical requirements into broader social aims.
For Chinese tech firms, this means aligning innovation efforts with regulatory expectations. Large domestic models like Ernie Bot and other generative AI products will likely be subject to more rigorous testing and oversight before public deployment, particularly in emotionally engaging contexts.
While some entrepreneurs argue that rigorous regulation could slow innovation, others suggest that clear standards can provide stability and competitive advantage by reducing uncertainty for developers and investors.
Global Context: China’s Governance Role in AI
China’s move to draft rules on human-like AI interactions resonates beyond its borders. As one of the world’s largest AI markets and a major hub for research and development, China’s policy choices influence global discussions about the future of AI governance.
In mid-2025, China’s leadership proposed the creation of an international organization for global AI cooperation, underscoring the country’s interest in shaping cross-border standards and frameworks for responsible AI development.
The debate over AI regulation is far from uniform worldwide. Different countries and regions emphasize varying priorities — from free speech and innovation in the U.S. to ethical and human rights concerns in Europe. China’s emphasis on psychological safety, national security, and content control represents a third approach that blends technological governance with state and social governance objectives.
What’s Next for China’s AI Rulemaking
The draft rules are now open for public comment, an important step in a regulatory process that will likely continue evolving in the coming months. Stakeholders including tech firms, civil society actors, and academic experts will weigh in before final versions are adopted.
If finalized, these rules could set new precedents not just for China but for how humane interaction and emotional impact are considered in AI governance globally. As AI systems become more embedded in everyday life — from chatbots to virtual companions — the question of how societies regulate their influence on human emotion, behavior, and decision-making becomes increasingly urgent.
China’s latest draft rules underline that AI regulation is no longer just about data and algorithms — it’s about the human experience those technologies shape.
