Stricter EU rules on AI remain uncertain after lobbying by major countries

The technology behind the ChatGPT chatbot and similar AI tools is increasingly likely to escape Europe’s strict AI rules, known as AI law. A dramatic change, of course: This summer, both EU member states and the European Parliament were in favor of this view.

This is because the EU’s three largest countries, Germany, France and Italy, are making a comeback. They fear that too strict rules will affect not only American tech giants but also European parties, and they want to prevent this. Among others, French artificial intelligence company Mistral AI is also interested in open positions.

Today, the last and important round of negotiations takes place between the European Parliament and the Council of the European Union (Member States), led by the European Commission.

Artificial intelligence law has been worked on for years. Basically, the greater the risks, the stricter the rules. The rules were already set before ChatGPT was launched at the end of last year.

Brussels wants to closely follow developments surrounding artificial intelligence. The basis for this are the so-called core models: for example, the core technology on which ChatGPT runs.

For example, developers should be required to ensure that their technologies do not harm human rights, health or the environment. They must also prevent misuse of copyrighted material.

Make your own rules

However, this plan is now being questioned. Last month, France, Germany and Italy published a joint document advocating allowing companies to set their own rules.

This means powerful tech companies like Microsoft, Google and Meta, as well as ChatGPT maker OpenAI, will themselves explain how their technology works and what it can and can’t do. There must be some kind of oversight.

Critics fear this will cause companies to focus primarily on their own interests and less on social interests.

“Then you don’t have a stick behind the door,” replies Catelijne Muller, president of the AI ​​organization ALLAI. She has been following the discussions about artificial intelligence law for years. “What they want, of course, is not to get any punishment.” This is included in the proposal of all three countries. They want to start without an enforcement system and only consider doing so after multiple violations have been committed.

for janjoker

The influence of the lobby document is great because the Member States concerned have great influence. “It seems like we’ve been talking about this for months,” sighs MP Kim van Sparrentak (GroenLinks), who is participating in the negotiations on behalf of the Greens.

According to him, there is an important technology lobby behind the document. Not just from the American tech giants, who are known to try to exert a lot of influence on all kinds of big tech issues in Europe, but also from two European players. Van Sparrentak points to French Mistral AI and German Aleph Alpha.

In particular, he mentions Mistral’s leading lobbyist, Cédric O, who was previously digital affairs minister in President Macron’s government. He was one of the driving forces behind a letter from European companies this summer expressing concerns about artificial intelligence legislation, according to EU news site Euractiv. EU Commissioner Breton pointed out in the business magazine La Tribune that Mistral does not serve the public interest.

When asked, Aleph Alpha confirmed that the three major countries supported his plan. “Due to dynamic development, a design that can adapt quickly is required.” In other words: They want more freedom. Mistral AI has not yet responded.

Speaking to Sifted, a Europe-focused technology site, Cédric O warned in late October that AI legislation could destroy his company, Mistral. He is particularly concerned about the administrative burden imposed by the law.

Mistral has twenty employees, has been in operation for only six months, and is said to already have a fortune of between 1 and 2 billion euros. The company received an investment of 105 million euros four weeks after its establishment.

Van Sparrentak, a member of the European Parliament, said that in the last round of negotiations the Parliament would try to set binding rules, such as testing of systems. If this is not successful, companies will be asked to “do their best” at most.

Source: NOS