European government leaders have agreed to a provisional version of the AI Act. The law still needs to be adopted by Parliament and the member states, but the provisional agreement marks an important step for AI regulation.
The Committee of Permanent Representatives, Coreper, has agreed to draft the AI Act. The AI Act is intended to regulate the rise of artificial intelligence. The law has not yet been approved by the European Council, but Coreper is doing the preparatory work for this, making it almost certain that the individual member states will also vote in favor of the bill.
With Coreper’s agreement, an important step has been taken for the AI Act. In recent months, some Member States, especially France, have been opposed to the law. France wants European companies to continue to innovate and specifically points to start-ups such as Mistral AI from that country. The AI Act would limit those companies too much in their options. Germany and Italy also had difficulty with parts of the law.
The provisional agreement stipulates, among other things, in which areas it is prohibited to use AI. This is, for example, the case for assigning risk scores, as the Netherlands used for years with various algorithms such as SyRI. It has also been established that ‘low risk’ applications are not covered by the law and AI chatbots must clearly indicate that artificial intelligence is being used.