As EU lawmakers head into the final weeks of negotiations for the bill, the association raised concerns over the handling of foundation models and general-purpose artificial intelligence (GPAI), arguing for the limitation of transparency models.

Digital Europe reports its concerns that the overregulation of foundation models and GPAI will lead to the stifling of European AI startups.

They said: “For Europe to become a global digital powerhouse, we need companies that can lead on AI innovation also using foundation models and GPAI. As European digital industry representatives, we see a huge opportunity in foundation models, and new innovative players emerging in this space, many of them born here in Europe. Let’s not regulate them out of existence before they get a chance to scale, or force them to leave.”

The association reported that only 8% of European companies currently use AI, and around 3% of the world’s AI unicorns come from the EU.

The signatories emphasised the cost of compliance for SMEs in the new legislation, with Digital Europe stating the price tag could be well over €300,000 for a startup.

Digital Europe said: “The AI Act does not have to regulate every new technology, and we strongly support the regulation’s original scope focusing on high-risk uses, not specific technologies.”

The letter also argued that the regulation would clash with existing sectoral regulation.

Signatories of this letter include Camilla Ley Valentin, CEO, DI Digital, Michał Kanownik, president, Digital Poland, Maria Shevchuk, acting СЕО, IT Ukraine Association, and Stella Morabito, director general, AFNUM.

The trade association concluded with these recommendations and comments:

• The risk-based approach must remain at the core of the AI Act.
• Regulatory flaws will be aggravated at the sectoral level, such as healthcare.
• Regulating GPAI and foundation models requires focusing on information sharing, cooperation and compliance support across the value chain.
• The EU’s comprehensive copyright protection and enforcement framework already contains provisions that can help address AI-related copyright issues.