AI Act: EU reaches agreement on rules for artificial intelligence

12/8/2023, 11:56:29 PM

Highlights: AI Act: EU reaches agreement on rules for artificial intelligence. After a long period of negotiations, the AI Act is now in place. There are to be stricter rules for the use of artificial intelligence in the EU. The hope is that the rules will be copied worldwide. The EU Commission proposed the law in April 2021. The higher the potential hazards of an application, the higher the requirements should be. The European Parliament and the states still have to approve the project that has now been agreed, but this is considered a formality.

After a long period of negotiations, the AI Act is now in place, with which the EU wants to be a pioneer worldwide. There are to be stricter rules for the use of artificial intelligence.

Enlarge image

EU headquarters in Brussels

Photo: Virginia Mayo / dpa

Stricter rules are to apply to the use of artificial intelligence (AI) in the EU in the future. Negotiators from the European Parliament and EU member states agreed on the corresponding rules on Friday evening in Brussels after long negotiations. According to the EU Parliament, this is the world's first AI law. EU Internal Market Commissioner Thierry Breton spoke of a "historic" agreement.

Artificial intelligence usually refers to applications based on machine learning, in which software sifts through large amounts of data for matches and draws conclusions from them. They are already being used in many areas. For example, such programs can evaluate CT images faster and with greater accuracy than humans. Self-driving cars also try to predict the behavior of other road users. And chatbots or automatic playlists from streaming services also work with AI.

The EU Commission proposed the law in April 2021. Accordingly, AI systems are to be divided into different risk groups. The higher the potential hazards of an application, the higher the requirements should be. The hope is that the rules will be copied worldwide.

Recently, however, the negotiations almost failed – due to the question of regulating so-called basic models. These are very powerful AI models that have been trained on a broad set of data. They can be the basis for many other applications. This includes, for example, ChatGPT. Germany, France and Italy had previously demanded that only specific applications of AI should be regulated, but not the basic technology itself. But the planned rules on facial recognition by AI, for example for national security purposes, also caused controversy. The German Digital and Transport Minister Volker Wissing (FDP) appealed to the EU to proceed in an internationally coordinated manner and "not to dare to go it alone".

The European Parliament and the states still have to approve the project that has now been agreed, but this is considered a formality.

max/dpa/AFP