Open letter: For uniform AI regulation in the EU

The EU risks missing out on the era of artificial intelligence, according to an open letter signed by numerous companies and researchers who work in the EU and develop and provide AI services. The issue at stake is which data may be used for the training of AI models. The signatories warn that if data from the EU does not flow into the most widely used models in the world, the models will also lack an understanding of the people, languages and cultures of the EU.

The letter appears to have been largely driven by Meta. Mark Zuckerberg himself is one of the signatories. Among other things, the letter states: "The EU will also miss out on other innovations, like Meta's AI assistant, which is on track to become the most used AI assistant in the world by the end of this year." Meta's AI models are open source, and Mark Zuckerberg has already stated several times that he believes this is the only viable path in the field of AI. According to Zuckerberg, an entire ecosystem will be able to develop around open source models.

Signatories include SAP, Spotify, Thyssenkrupp, EssilorLuxottica, Artefact and several scientists as well as civil society and trade associations.

Criticism of the GDPR not the AI Act

However, the criticism is not, as one might initially think, directed at the AI Act, i.e. the regulatory framework for AI that only recently came into force. The open letter is primarily about data protection. The Irish Data Protection Authority (DPC), which is responsible for Meta in the EU, among others, had prohibited the company from using the data of people who are on Facebook and Instagram for AI training. Meta had wanted to opt out to ensure that they could use all posts, but not private messages. Although the DPC initially allowed the opt-out procedure, it reversed this decision after protests. Meta stopped the use of the data.

In fact, it can be assumed that all other providers of large AI models have already used the public posts on the platforms to train their models – without consent. Mira Murati, CTO of OpenAI, for example, confirmed in an interview that they have used everything that is publicly available. Whether this is legal has yet to be decided. This applies to social media posts as well as YouTube videos, newspaper articles and works of art. However, the latter is a matter of copyright law and not data protection.

According to the open letter, the EU is at risk of losing touch and weakening economic growth. "Studies assume that generative AI could increase global GDP by ten percent over the next ten years, and EU citizens should not be deprived of this growth."

To avoid this, the signatories call for: "harmonized, coherent, swift and clear decisions under EU data protection rules that allow European data to be used in AI training for the benefit of Europeans."

(emw)