China's AI Personification Regulation Is Here: World's First Specialized Law, Effective July 15
On April 10th, five government departments—Cyberspace Administration, NDRC, MIIT, MPS, and SAMR—jointly released a document: “Interim Measures for Managing AI Personified Interactive Services,” effective July 15th.
This arrived faster than I expected. While everyone was discussing “should AI be regulated” and “how to regulate,” we got the “world’s first specialized regulation.”
What Is “Personified Interactive Service”?
First, let me explain this concept—it’s crucial.
“Personified interactive service” refers to: AI systems that interact with users by simulating human appearance, voice, expressions, and movements. Simply put, those “human-like” AIs—virtual anchors, AI companions, personified customer service, AI education assistants, etc.
Regulation in this area was previously blank. AI chatbots were “information services,” AI-generated content was “content moderation,” but “AI pretending to be human interacting with people” had no specialized regulations.
This new regulation fills that gap.
Key Points of the New Regulation
I quickly scanned the document—several key points deserve attention:
First, defining “boundaries.” Personified AI cannot “pretend to be exactly like real humans”—it must prominently display “I am AI.” This requirement is quite reasonable—some AI companions and virtual partners became so realistic that many couldn’t distinguish real from fake, creating ethical risks.
Second, protecting “vulnerable groups.” The regulation specifically mentions “childcare support” and “elderly care services”—stricter regulatory standards for AI used by children and elderly. This aligns with my judgment: AI regulation focus should be on protecting those most vulnerable.
Third, data security and personal privacy. Personified AI involves massive user interaction data—how to store, use, and protect this data has clear requirements. Particularly the “cannot collect sensitive information without user consent” clause directly addresses many AI products’ pain points.
Fourth, service provider responsibility. If AI causes problems—generating harmful content, misleading users—the service provider bears responsibility. This means AI companies can’t just focus on technology without considering consequences.
World’s First—What Does It Mean?
Honestly, this “world’s first” title is both honor and pressure.
Honor in that China is leading in AI regulation again. Previously we had “Generative AI Service Management Measures,” now “Personified Service Management Measures”—the regulatory framework keeps improving.
Pressure in that regulation implementation is a massive project. Personified AI has too many scenarios—from virtual anchors to AI education, from smart customer service to AI companions, each with unique characteristics. How to control without stifling innovation is the challenge regulators face.
Impact on AI Industry
I predict several changes:
First, AI products must “de-camouflage.” Previously many products pursued “the more human-like the better,” now it’s reversed—users must immediately recognize “this is AI.” This may force product design changes.
Second, compliance costs rise. AI companies need dedicated compliance teams ensuring products meet new requirements. This increases operational costs but also raises industry barriers.
Third, innovation direction adjustment. Can’t rely on “acting human” to attract users anymore—AI products must compete on genuine functionality—solving problems, improving efficiency. This benefits long-term industry development.
My Personal Take
Honestly, I’m cautiously optimistic about this regulation.
Optimistic because the regulatory starting point is right—protecting users, standardizing industry, preventing risks. AI technology develops too fast—without rule constraints, problems easily arise.
Cautious because implementation has many uncertainties. Regulatory text is static, but AI applications are dynamic. How to balance “regulation” and “innovation” requires continuous adjustment in practice.
But at least we have a starting point. The world’s first specialized regulation means China is exploring a new path in AI regulation. This path won’t be smooth, but at least it’s begun.
July 15th implementation gives 3 months. AI companies, time to prepare.
Note: This article analyzes based on the publicly available “Interim Measures for Managing AI Personified Interactive Services” document and does not constitute legal advice.