Recently, the X platform began translating all posts (written in various languages) into the native language of the reader.
Thanks to this new feature, a few days ago I was able to read a post written by Luiza Jarovsky, which discussed the new Chinese law on artificial intelligence.
Thanks to the article published by Luiza Jarovsky in her newsletter, I better understood the main aspects of the law. For those who want to read the article, this is the link.
China is the first country in the world to require AI companies to prevent the risk of AI anthropomorphism to protect people.
The law is called Interim Measures for the Management of Anthropomorphic AI Interaction Services. It was published on April 10, 2026, and will come into effect on July 15, 2026. As of April 2026, companies have three months to comply with the new regulation.
The main points of the law (from the final version published by Chinese authorities) are these: the regulation requires companies to take security responsibility for the entire service lifecycle, from initial development to shutdown. They must also introduce concrete measures against emotional dependence, psychological manipulation, and replacement of real human relationships.
The rules for minors are very strict: it is forbidden to offer “virtual companions” or “virtual relatives,” a protected mode is mandatory, parental consent is required under 14 years of age, and content that could encourage dangerous behaviors, provoke extreme emotions, or lead to bad habits is prohibited.
The law also mandates clear labeling: users must always be warned that they are speaking with artificial intelligence and not a real person. Conversations cannot be used to train models without separate and specific consent.
Finally, the entire regulation is based on the Chinese principle of “intelligence towards good,” which seeks to combine technological development with societal protection.
The European Union adopted the AI Act (the Artificial Intelligence Regulation) on March 13, 2024. The regulation came into force on August 1, 2024, and foresees gradual application starting in 2025. In the United States, there is still no specific federal law, but some states are moving: New York in November 2025 and California in January 2026 have approved rules dedicated to “chatbot companions,” with transparency obligations, protocols to prevent self-harm, and greater protections for minors.
Despite the efforts of European and American lawmakers to protect people’s health, compared to the Chinese regulation, there is still a long way to go.
The Chinese law, which will come into force in one of the most restrictive countries in the world regarding individual freedoms, shows—albeit in a clearly paternalistic way—that the state truly cares about the mental health of its citizens. At least this is what emerges from reading the text.
In Europe, instead, we have left too much space for Big Tech self-regulation, allowing them to operate almost undisturbed. We intervene only in exceptional cases, as recently happened with Meta.
However, one undeniable fact remains: the country that approved this law is the same one that created the Social Credit system. I do not expect major openings on individual freedoms, yet on this specific point they have shown they understand very well the risk of AI anthropomorphism.
I observe, however, that in Europe many regulations (AI Act, DSA, monitoring cameras in new cars, etc.) can be read in two ways: as tools for protection or as forms of increasingly pervasive control over our daily lives. In short, a soft Social Credit, under another name.
In the technological challenge, the Chinese are clearly ahead of us, and not only on the topic of AI anthropomorphism. In Europe, instead, there is often an almost childish enthusiasm for technology, or commissions and “Algoethics” task forces are created both in institutions and companies. Honestly, I see them as much ado about nothing: there is a lot of talk about ethics, but rarely do they address the real risks.
It is true that in Europe, besides often facilitating Big Tech, there is also a real concern for fundamental rights such as non-discrimination, privacy, and human dignity. This is an important point that distinguishes the European approach from the Chinese one, where everything is subordinated to the stability of the Party and the State.
In the end, however, one point remains clear: without even minimal IT knowledge among politicians and managers, we Europeans will continue to lose technological challenges both against China and the United States.
I am also truly surprised that intermediary bodies—starting with trade unions—are completely absent on these issues. Perhaps they do not realize how many millions of jobs, precisely in the field of artificial intelligence, will be lost in the coming years.
Recent experiments conducted in war scenarios clearly demonstrate how dangerous it is to rely completely on AI.
I leave you with a question I have no answer to: are we really sure we want to hand everything over to AI without putting up any resistance?
