A coalition of Europe’s biggest enterprises, including Airbus SE, Mercedes-Benz AG and BNP Paribas Group is asking the European Union to freeze the implementation of a new artificial intelligence law, saying it could prevent the bloc from keeping up with the pace of development in China and the United States.

The chief executive officers of 44 companies, which include Europe’s top AI developer Mistral AI, have asked European Commission President Ursula von der Leyen to delay the AI Act for two years. They say its unclear, overlapping rules will discourage European investment in AI and slow down innovation, the Financial Times reported.

In an open letter to EU officials, the company heads argue that the act will put Europe’s AI ambitions at risk, because it “jeopardizes” the development of European champions and hampers their ability to deploy the technology at the scale needed to compete globally.

Among the business leaders who added their signatures to the letter are the CEOs of the French retailer Carrefour Group S.A., the Dutch electronics firm Koninklijke Philips N.V. and chipmaking equipment manufacturer ASML Holdings N.V.

The companies have strong support in their efforts to pause the rollout of the legislation, with the U.S. government, big technology firms and European business groups all putting pressure on the EU in recent weeks. On Wednesday, Brussels hosted representatives from a number of U.S. tech companies that came to participate in the creation of a revised, software version of the AI Act.

Those discussions focused on a “code of practice” within the Act that’s designed to help guide companies on compliance when deploying large language models such as OpenAI’s GPT-4 and Meta Platforms Inc.’s Llama. The code was originally set to be published in May, but has since been delayed and is expected to be watered-down.

On Monday, the EU’s technology commissioner Henna Virkkunen promised that the guidance would be published before the AI Act formally comes into effect, in August. “We will publish the code of practice before that to support our industry and SMEs to comply with our AI Act,” she stated.

It’s thought that EU Commission officials and member states have been holding intense negotiations behind closed doors in order to try and temper some of the new guidelines and simplify the act’s staggered rollout timeline.

Though the AI Act officially came into law in August 2024, a number of the most important measures within it will only take effect later this year, or beyond. That has resulted in a lot of confusion among European businesses, the Financial Times reported.

“It’s a classic example of regulation that doesn’t take into account the most important thing for industry, which is certainty,” said Cooley LLP co-chair Patrick Van Eecke.

The letter was sponsored by an initiative called the European AI Champions, which is a coalition of 110 companies spread across multiple industries. The group is calling for a two-year pause on the AI Act’s rollout, arguing that to do so would “send a strong signal that Europe is serious about its simplification and competitiveness agenda.”

They’re joined by the founders of and investors in numerous European startups that have voiced similar concerns over the AI Act. Earlier this week, the heads of more than 30 startups signed a letter that said the legislation is a “rushed ticking time bomb,” adding that they believe its unclear rules on the use of general-purpose models could result in a patchwork of national regulations. They fear this would give better-funded U.S. companies a significant competitive advantage, and damage Europe’s own nascent AI industry.

Many European business chiefs fear that using LLMs could leave them at the mercy of the same regulations that govern the biggest technology firms, and are especially worried about being held liable for copyright infringements. The lack of clarity on how EU member states will apply the new rules may deter the adoption of AI tools, the CEOs say, putting them at a disadvantage against their competitors in Asia and the U.S.

It’s a problematic issue, because lawmakers must be careful not too harm development while still providing safeguards against the risks of “out-of-control” AI systems, said Rob Enderle of the Enderle Group.

“We need a close partnership between big business and the government to strike a balance and find the right path, but it may prove impossible to find at a time when companies are fighting against any kind of control, and politicians are more interested in looking like they’re doing something, as opposed to taking their time to do the right thing,” the analyst said.

According to Enderle, there is a real sense of urgency too, as European companies believe they’re already falling behind China, where enterprises have very few restrictions on how they’re using AI. At the same time, he believes governments still don’t fully understand AI or the risks it presents.

“This probably won’t end very well either for the companies or the governments concerned, and most definitely not for the citizens those governments are trying to protect,” Enderle said. “AI is a massive force multiplier, and it can do tons of harm if it’s not properly controlled. But if it’s too aggressively throttled, those companies that need it to compete will likely fall behind their Asian rivals, and that will do massive economic damage to those countries doing the throttling.”

Holger Mueller of Constellation Research Inc. said the EU had good intentions in rushing to approve its AI legislation, but it’s clearly going to have some unintended consequences that will hinder adoption, and any efforts to correct those issues will likely be too slow to be of any use.

“That’s why these blue chip companies are trying to pause the EU AI Act before it comes into force properly,” Mueller said. “At issue is the EU’s federal patchwork of rules in different member states, which will make AI adoption difficult due to regulatory overburdening. But if the Act is delayed, it will likely need to be replaced by entirely new regulation, because the speed of innovation means it’s unlikely to have much applicability in 2027 and 2028.”

For its part, the EU Commission’s only response so far has been to insist that it’s “fully committed to the main goals of the AI Act,” which calls for the establishment of harmonized, risk-based rules across the EU.

Image: SiliconANGLE/Dreamina