The European Parliament adopted today a regulation on artificial intelligence, the aim of which is to guarantee the safety of citizens and respect for their basic rights, while encouraging innovation.
The members of the European Union reached an agreement on the proposed regulation in December 2023, and 523 MEPs have now voted for it, 46 were against it, and 49 abstained, the Parliament in Strasbourg announced.
The regulation will completely prohibit, or only to a certain extent allow, the use of dangerous artificial intelligence systems depending on the degree of risk they represent.
The new rules prohibit the use of artificial intelligence applications that threaten the rights of citizens, such as biometric categorization systems based on certain personal characteristics or appearance - so-called profiling.
The creation of facial recognition databases based on random collection of images from the Internet will also be prohibited. In the workplace and in schools, it will not be allowed to use AI applications to "read" the emotions and moods of employees or students.
It will also be prohibited to assess the social behavior of people with the help of AI, as well as to make forecasts based on profiling people according to their personal characteristics. AI will also not be used to manipulate human behavior or exploit human weaknesses.
In the area of enforcement, AI will be allowed in very clearly defined and, as stated, few situations. This use, according to the new directive, will be possible with a prior court decision and geographically and temporally limited to specific cases, for example during a targeted search for missing persons or to prevent a terrorist attack.
The use of AI in the enforcement of regulations after an event will be possible only with a court decision related to a specific criminal offense.
High-risk artificial intelligence systems will be subject to very clear obligations, due to their potential to threaten health, safety, human rights, democracy or the rule of law. Examples of such high-risk AI technology are those that will be used in critical infrastructure, education, and vital private or public services, such as healthcare or banking.
The category of high-risk systems also includes those that can be applied in the enforcement of regulations, border management, in the judiciary or, for example, in democratic processes, such as elections that can be influenced by AI. They will have to be extremely transparent and reliable and only be used under human supervision.
In addition, "Deepfakes" - fake photos, videos and audio recordings that are created and manipulated with the help of AI - will have to be clearly labeled as such.
Citizens will have the right to file complaints against AI systems and to have their rights explained about decisions related to such risky systems.
The goal of the regulation is to protect fundamental rights, democracy, the rule of law and environmental sustainability throughout the EU territory from AI systems that are considered risky, but also to strengthen innovation and Europe's leadership role in that area.
The adopted regulation should go through a check of legal terminology and be formally approved by the Council of the EU, and it should enter into force before the European elections, whereby EU members will have a period of six to 36 months for the implementation of certain provisions.
Bonus video:
