The Council of the European Union on Tuesday approved the Artificial Intelligence Act, the first such law in the world that could set a global standard for the regulation of artificial intelligence. The goal of the new law is to encourage the development and acceptance of safe and reliable artificial intelligence systems on the EU single market. At the same time, its goal is to ensure respect for the fundamental rights of EU citizens and to encourage investment and innovation in the field of artificial intelligence in Europe.
The Law on Artificial Intelligence applies only to areas covered by EU legislation and provides for exemptions such as systems used exclusively for military and defense purposes and for research purposes.
The law follows a “risk-based” approach, which means that the rules are stricter the greater the risk of harm to society.
“This law, the first of its kind in the world, is a turning point for the EU, and it relates to global technological challenges that create opportunities for our societies and economies,” said the Belgian State Secretary for Digitization, Mathieu Michel.
The Council’s decision came after the same law was voted in the European Parliament on March 13.
The new law should be published in the Official Journal of the EU in the next few days, and will enter into force 20 days after publication. The regulation will start to apply two years after its entry into force, with some exceptions for certain provisions of the law.
The new law differentiates between four types of artificial intelligence according to the risk they carry. Artificial intelligence systems with very limited risk, such as video games or spam filters, will have to meet only mild transparency regulations, while high-risk artificial intelligence systems will have to meet a series of requirements and obligations in order to be approved on the market; of the EU.
The law prohibits artificial intelligence systems that socially value and classify individuals into certain categories – by gender, gender identity, race, ethnic origin, migration status, citizenship, political belief, sexual orientation, religion, and the like. .
Unacceptable risk artificial intelligence includes storing photos of faces from the Internet or CCTV recordings, emotion recognition in the workplace and in schools, social scoring, predictive police action based exclusively on profiling a person. Citizens will be able to file complaints against AI systems that threaten their rights.
In order to ensure the correct implementation of the new law, the establishment of several bodies such as an office for artificial intelligence within the Commission, a scientific panel of independent experts to support implementation activities and others is foreseen.