BRUSSELS — The EU is flipping its script on artificial intelligence amid a global race to win cash and influence.

The European Commission is on Wednesday expected to postpone the implementation of landmark AI restrictions by at least a year as part of sweeping changes to digital rules aimed at staying competitive with the U.S. and China.

For years EU policymakers focused on making regulations to ensure the technology can be trusted. Now, in a year that saw major advances in artificial intelligence and Donald Trump reenter office, the EU is letting go of its dream of being the global leader on regulating AI.

The Artificial Intelligence Act, which took years to negotiate, is not even fully in place yet. Throughout 2025 a growing chorus of national governments and executives from tech companies and industry lobby groups have called for a delay of a part of the law, putting the issue at the center of a wider fight in Brussels over how the EU should balance regulation and innovation.

Wednesday’s proposal will see industry voices win out, with the announcement made under the same Commission president that heralded the original law as a “historic moment” to make people safer.

While the EU executive will present the proposal as a technical adjustment that will ultimately make the EU’s regulation more effective — on the basis that changes will help industry to comply — it follows an intense lobbying effort by the Trump administration in Washington and from corporate lobbies in Brussels against the bloc’s digital rules.

“A part of the message that Europe is giving to the rest of the world is that it is open to pressure from tech companies and other nations,” said Natali Helberger, a professor of law and digital technology at the University of Amsterdam. “I would say this harms the credibility.”

Under the plans expected Wednesday, a series of AI practices that are classified as high risk — for example using artificial intelligence in recruitment, to assess people’s suitability to get loans or to score exams — won’t face obligations for at least a year longer than planned.

A big part of the justification for the decision has been concerns that the regulations will prevent Europe from being competitive at a time when it needs to level up. Tech lobbies have slammed the foreseen timeline as “unworkable.”

“If we only could take the foot off the brake and give innovation a bit more chance, I think that’s all we need,” Germany’s Digital Minister Karsten Wildberger said Tuesday when asked about the Commission’s upcoming proposal.

The plans are prompting pushback from civil society.

“The Commission seems intent on destroying fundamental rights safeguards and setting us up for months, if not years of infighting and legal uncertainty without any tangible gains for EU competitiveness,” said Daniel Leufer, senior policy analyst at AccessNow.

Other changes expected Wednesday would exempt more companies from certain rules altogether, and would also give industry a grace period on new rules for watermarking visual content made by AI.

Too ambitious?

The bloc’s AI rulebook was adopted in August 2024 but the rules were always intended to take effect gradually.

Some AI practices that carry an “unacceptable risk” such as predictive policing or social scoring have been forbidden since February. The most complex AI models, such as OpenAI’s GPT, have also had to play by a separate set of rules since August.

The rules that the EU executive is now pressing pause on — those that pose a risk to people’s health, safety or fundamental rights — were slated to take effect in August next year.

Countries and companies argued a delay was necessary due to a delay in the technical standards, designed to help companies comply with the requirements. Standardization bodies missed the deadline to deliver on them twice, and now the standards won’t be ready until 2026.

The timeline to come up with standards was a “bit ambitious from the start,” a representative from the standardization bodies told POLITICO in September.

By branding it as a technical delay due to the lack of guidance, some in favor of a pause are choosing not to label it as a retreat, but instead to suggest a little more time is needed to get things right.

“Many companies would welcome this,” said Wildberger. “But equally important is that we use the time to get certain things right. It’s not just: we postpone it. No, we have some work to do.”

Germany and France came out publicly in favor of a one-year pause on Tuesday. Sweden, Poland, the Czech Republic and Denmark all called for a pause or a grace period before.

Countries had a stake in delaying the process. “It is also motivated by the fact that so far, a lot of member states haven’t assigned and equipped their national regulatory authorities that must enforce the AI Act,” said Helberger.

Hitting pause “will give them more time to get their act together at the national level,” she said.

Wednesday’s proposal will need approval from EU countries and by the European Parliament before becoming final. There’s a hard deadline of August 2026 when the rules were set to apply.

Within Parliament, even critics of the pause have privately conceded defeat and are now focused on keeping the delay as short as possible and avoiding further pushback.

“Unfortunately, a pause now seems inevitable given the delay in developing the standards,” Irish Renew lawmaker Michael McNamara said last week after POLITICO first reported that the rules would be delayed by at least a year.

McNamara warned that there should be “no further delays,” because “if there were, it would undermine regulation and rule of law beyond just the AI Act.”

Mathieu Pollet contributed to this report.