The EU’s landmark AI Act would hurt smaller European companies and only benefit big players who can cover the high costs of compliance, the US warns.
According to documents seen by Bloomberg, the State Department’s analysis of the upcoming legislation is sounding the alarm on the European Parliament’s version of the act. Of particular worry are the rules surrounding large language models (LLMs), which form the base of most generative AI products.
The analysis found a number of these rules “vague or undefined.” It also raised concerns over the act’s focus on the development risks involved in AI models, rather than the risks related to their use.
Washington warned that the regulation could reduce productivity, cause job migration, and discourage investment in R&D and commercialisation within the bloc, tampering with the competitiveness of European companies.
The US feedback has already been shared with EU heads, people familiar with the matter told the newspaper.
Meanwhile, European firms have also expressed similar fears. In June, for example, executives from some of the bloc’s biggest companies expressed “serious concerns” in a letter sent to the Parliament, Commission, and member states.
“The draft legislation would jeopardise Europe’s competitiveness and technological sovereignty without effectively tackling the challenges we are and will be facing,” said the letter.
In contrast, some EU countries including Italy have started regulating generative AI even before the act’s enforcement, while a July survey showed that European consumers believe that the technology should be heavily regulated.
The AI Act will come into force in late 2025 or early 2026. Amid resistance from the business sector and fears of extinction, it will have to walk the fine line between ensuring safety and boosting innovation.