01.01.2026, Moscow.
While the West is debating who is to blame if so-called artificial intelligence (AI) transfers a user’s money to fraudsters, drives them into depression, or even to suicide, China is moving on to concrete steps, Rossa Primavera News Agency IT Desk notes.
On December 27, 2025, the Cyberspace Administration of China published a draft of the “Provisional Measures on the Administration of Human-like Interactive Artificial Intelligence Services” for public discussion. The restrictions concern technologies that imitate human activity — thinking or communication via text, images, audio, or video.
Responsibility is placed on service providers. They are required to create management systems, including algorithm verification, ethical review, content moderation, and the protection of data and personal information.
The rules prohibit the generation or dissemination of content that harms state security, violates public order, incites hatred or religious conflicts. The creation of pornography, content promoting violence, or content that insults or defames others is also prohibited. The use of algorithms to manipulate users, mislead them, create emotional traps that push them toward irrational decisions, or extract confidential information is forbidden.
For minors, providers are required to create a special mode, obtain parental consent, set time limits, and provide parents with monitoring functions. The use of minors’ data to train models without separate parental consent is prohibited.
For elderly users, it is forbidden to create services that imitate their relatives or close acquaintances. If extreme emotions or suicidal tendencies are detected in users, the system must issue template responses recommending that they seek professional help. In obvious cases, the dialogue must be switched to a human operator to contact relatives or emergency services.
Providers are required to clearly indicate that the user is interacting with AI. If a service is used continuously for more than two hours, a dynamic reminder to take a break must be displayed. Providers must ensure that users have a simple way to exit emotional support modes. They are also required to use training data that correspond to socialist values and reflect the best traditions of Chinese culture.
The date for the introduction of the new measures has not yet been announced; authorities will accept public comments until January 25. However, the law is planned to be introduced as early as next year.
Thus, instead of reacting to AI-related scandals, China has decided to act preemptively. It is building a legal and technical barrier in advance against the most obvious and dangerous abuses of the new technology. Beijing clearly understands that no real or even expected progress matters if the human being or statehood is lost.
It is worth noting separately the requirement to train models on data reflecting socialist values and the best traditions of Chinese culture. This is an attempt not merely to counter the erosion of national identity, but to use the technology to strengthen it.
Of course, it is still difficult to predict how this will be implemented in practice. But it can be said unequivocally that such an approach is better than blind reverence for AI and the desire to introduce it into all spheres of life.
Source: Rossa Primavera News Agency
