We are talking about chatbots such as ChatGPT and Gemini, which can be used for text or image generation as well as programming.
Under new EU rules, AI developers are required to disclose how their models work and the data used to train them. The most advanced models, considered potential sources of risk to the public, are required to document the security measures taken. The innovations are aimed, among other things, at protecting intellectual property and the rights of authors and content owners.
Individuals can sue AI for such violations as early as August 2. The new European AI Office will start checking new models from August 2026, and models released before August 2, 2025 – from August 2027. Experts have estimated that the amount of possible fines imposed could reach $17.1 million (about 14.75 million euros), which is 3 percent of the annual AI services market.
It should be noted that the world’s first international AI law was approved by the European Parliament on March 13, and by the relevant ministers of the EU member states on May 21, 2024.
Thus, on February 2, 2025, after a transition period, the first rules for the use of AI in the European Union came into force. In particular, it was forbidden to use artificial intelligence to create a social rating of EU citizens (such technology is used in China). Also in the EU, it is not allowed to recognize emotions at workplaces or in educational institutions.
In addition, the EU prohibits the manipulation of human behavior with the help of AI and the exploitation of the vulnerability of individual citizens or social groups. This includes, for example, “robotic calls”, which can be used for fraud.
The recognition by artificial intelligence of people’s faces in public places using CCTV cameras is also prohibited in the EU. An exception is made only for law enforcement agencies.
