New rules for handling artificial intelligence (AI) have come into force in the European Union. According to the EU AI Act, AI systems that pose "unacceptable risks" to safety, health or fundamental rights are prohibited from this day on, with the exception of causes related to "national security". The world's first such regulatory act classifies all AI-based systems by risk level - from low to unacceptable. AI systems with a high level of risk - for example, in the field of critical infrastructure, employment, healthcare or banking - must meet a number of requirements to be allowed on the EU market. For low-risk applications, only limited transparency and information obligations apply.

An unacceptable risk is posed by artificial intelligence systems that can be used to deliberately influence people's behavior and thus manipulate them. The ban applies to them, as well as to "social scoring" based on artificial intelligence, that is, scoring in accordance with the desired behavior. Manipulative AI systems that use deceptive technologies to influence people's behavior include, for example, voice toys that encourage children to dangerous behavior.

The same category includes automated "robotic calls" (Robo-Calls), which can be used to defraud elderly people. Facial recognition in public places, including through surveillance cameras, will also be prohibited in principle, but exceptions are provided: the police and other law enforcement agencies will be able to use this technology to investigate crimes related to human trafficking and terrorism. From February 2, companies developing or using AI will have to assess the risk level of their systems and take the necessary measures to bring them into compliance with legal requirements. According to the European Commission, the goal of the new regulation is not only to protect consumer rights, but also to ensure the proper use of AI. Suppliers and operators of AI systems must now ensure that all persons involved in the development or operation of such systems have a sufficient level of "AI competence".

The European Artificial Intelligence Regulation is the world's first comprehensive set of rules on artificial intelligence. In a world with a rapidly unfolding revolution in the field of artificial intelligence, it is designed to ensure a balance between innovation and protection against the risks that they may carry.

ORIENT