The European Parliament today adopted a decision to implement the AI ​​Law, the first regulatory framework for artificial intelligence. The MEPs are striving for a healthy balance between protecting fundamental rights and promoting innovation, reports the Belga news agency.

The much-discussed AI law is therefore being interpreted in concrete terms and there is no room for automatic face recognition. However, the European People’s Party (EPP), the hemisphere’s largest group in Strasbourg, made desperate efforts to allow such recognition in exceptional cases. The European Parliament, on the other hand, favors separate rules for generative AI, such as ChatGPT, Bard and DALL-E.

Historical agreement

The positions adopted today by the European Parliament are a world first, although they still have to be approved by the various Member States. Broadly speaking, MEPs agreed on a comprehensive set of rules to address the risks of AI while encouraging innovation while respecting European values ​​and standards in the area of ​​privacy and data protection.

The text is therefore based on existing rules on (product) safety and will primarily provide controls for companies. The position received an overwhelming majority of votes. 499 MPs voted in favor, only 28 against. 93 members abstained. The use of a given AI application is scaled based on the risk involved. If the risk is so high that it is unacceptable, as in Chinese social scoring, then it is forbidden altogether.

face recognition

Based on the Chinese model: According to the EP, automatic face recognition by AI is unacceptable. However, the largest group, the European People’s Party (EPP), wanted some exceptions. For example, they wanted to enable facial recognition in emergency situations, such as when searching for missing children or terrorist suspects. The European Council has previously approved the use of such identification systems. So let’s talk a little more about that.

In order not to hinder AI innovation, the European Parliament has granted certain exemptions for research activities and open-source AI components. The use of state-created controlled environments for testing artificial intelligence is also permitted. The AI ​​law will come into force in 2026 at the earliest. A final text has yet to be negotiated with all Member States. Therefore, there is a very good chance that the legislation will already be obsolete before it comes into force.