State Secretary Van Huffelen pleased with European AI agreement

State Secretary Alexandra van Huffelen is "extremely happy" with a European agreement on artificial intelligence (AI). Van Huffelen said this in an initial response to the provisional agreement between the EU countries and the European Parliament on rules for AI. "However, this does not release us from the obligation to continue examining the opportunities and risks of using AI technology and to enforce the rules," Van Huffelen continued.

The State Secretary emphasized the importance of safe AI systems for Europeans. On X, she wrote that "All Europeans must be able to trust that AI systems are safe and do not discriminate".

The responsible EU Commissioner Thierry Breton reported around midnight on Friday that a new European AI law had been agreed upon after days of negotiations. Among other things, this law classifies AI applications according to their risk. Roughly speaking, the more danger an application poses, the stricter the regulations. The most dangerous applications, such as Chinese-style citizen monitoring systems that reward or punish citizens for their behavior, will be banned. The same applies to technologies designed to manipulate people's behavior and detect moods and emotions, for example in the workplace or at school. The untargeted collection of images of faces to build up a recognition database will also be taboo.

The agreement still has to be submitted to the EU Parliament and the EU member states, but this is usually a formality. Breton spoke of a "historical" moment.

"Dealing with AI is about finding the right balance between exploiting the opportunities and the concerns," said the outgoing State Secretary. "There are many sectors in which the Netherlands is strong and in which AI will play a major role. Think of agriculture, education, healthcare, peace and security. The Netherlands is working internationally to ensure that we can trust this technology. For example, to prevent discrimination in the use of AI to grant certificates or transparency of algorithms in the financial sector."

According to the Euro Commissioner, the EU is not only slowing down revolutionary technologies, but is actually spurring on young European companies and researchers. The two most powerful EU member states, including Germany and France, recently warned that overly strict rules could kill innovation. The EU could then miss the boat and, as with other digital revolutions, fall behind the USA and China.

Widely used AI applications such as ChatGPT must disclose how they work. They must pay attention to the copyright of the texts and other data they use to "learn". Developers of the most powerful models must also test how they behave when malicious actors try to manipulate them and must report incidents.

The negotiators argued for a long time about artificial intelligence in facial recognition, which the EU Parliament wanted to abolish, while the Member States wanted to allow it if it served security purposes. Now they are being given room for it, albeit with conditions and restrictions.

The new rules could all come into effect in 2026 if they are officially adopted early next year. The bans will come into force after just six months, and the transparency rules after one year, among other things. Violations could result in heavy fines.