EU lawmakers are said to be considering pausing the rollout of the AI Act following months of tense negotiations, confusion over how the law will be implemented and backlash from industry players at home and abroad.
According to a recent report from specialist legal publication MLex, the European Commission is considering a move to pause enforcement of the Act and introduce amendments to “simplify” the law. It comes as the Trump administration increasingly pushes back against the rules, which it sees as too weighty and unfairly targeting US companies, which dominate the global tech industry.
The AI Act, which came into force last year, is the most comprehensive set of laws governing the technology in the world. It introduces a long list of new rules for organisations using the tech in the EU, with an emphasis on ethics, safety and transparency.
The legislation took years to negotiate and encountered strong opposition from some European AI businesses. As a result, proposals to simplify the rules have been welcomed by some.
“Any effort that aims to improve the regulatory environment or reduce bureaucracy in Europe is a good thing,” Jonas Andrulis, the founder of German AI darling Aleph Alpha, tells Sifted.
At the same time businesses have been preparing for months for implementation of the rules, which is scheduled to happen gradually until 2027. Observers say reopening the text now for potential changes would cause significant uncertainty.
“We’ve done the legislative work of finding a compromise everyone agrees with and businesses have integrated the new constraints for over a year now,” says Marianne Tordeux-Bitker, director of public affairs at startup and VC lobby France Digitale.
“This creates total uncertainty… It is completely contemptuous of businesses.”
Stopping the clock
The report says the European Commission is considering a ‘stop the clock’ move to delay enforcement of the Act and pass amendments to simplify some of its rules. The Commission did not reply to Sifted’s request for comment.
A key matter of controversy has been around enforcement of the Act for general-purpose AI (GPAI) models, which is set to begin in August, and concerns powerful generative AI models such as those developed by OpenAI, Meta, Anthropic and French startup Mistral AI.
To help GPAI providers comply with the Act the EU has been working on a voluntary code of practice, which contains guidelines on the regulation of general-purpose AI models like ChatGPT and others, with guidance around for transparency and copyright-related rules would be expected to follow. Following the most recent draft published in March, a final version of the code is now expected after the summer.
The US government reportedly sent a letter last month to the European Commission arguing the code is too burdensome and shouldn’t be adopted in its current form — and US-based Big Tech companies like Google and Meta have also repeatedly spoken out against the AI Act and the code of practice.
Kai Zenner, a digital policy expert and head of office for lawmaker Axel Voss, who helped negotiate the text of the AI Act, says delays to the code of practice and wider AI Act rollout may be positive for businesses. “It’s good they’re taking a little more time, the whole process was too rushed,” he says.
“I’m sympathetic to the idea of adding another year to the timeline,” Zenner says of the reported pause. “But I don’t think we should rerun the debate over the AI Act. Reopening it, questioning whether or not it’s going to be enforced, it’s creating a lot of uncertainty for smaller market players. I think it’s really toxic for them.”
For now it is unclear what exactly the Commission might be considering to change in the AI Act.
“The uncertainty is the big issue and will weigh on founders, in terms of time and cost,” Rasmus Rothe, cofounder of German AI VC Merantix Capital, tells Sifted.
But he adds the signal could be positive. “The larger perception among founders is that European legislators are finally hearing the calls that we need to focus on innovation over regulation,” says Rothe. “The sense that the AI Act might in effect be watered down contributes to that positive, innovation-first notion.”
Pressure from the US
Some in EU policy circles have vocally rejected bowing to pressure from the US.
“Are we really going to backtrack on our foundation of European law just because a few companies don’t agree with it?” says France Digitale’s Tordeux-Bitker. “If this is a reaction to the Trump administration, it is a political mistake.”
“It’s important to keep the essential principles that are the product of a compromise that has been incredibly difficult to reach,” Laurent Daudet, cofounder of Paris-based generative AI company LightOn, tells Sifted. “We shouldn’t undo the law under the pretence that we are leaning towards deregulation because of the Trump administration.”
Meanwhile, some argue pausing the law could be a welcome relief for AI startups given that details regarding implementation are still being ironed out.
“Central compliance tools and guidelines to be implemented under the AI Act… are not ready yet,” Robert Kilian, CEO of German AI testing provider CertifAI and a board member of the German AI Association, tells Sifted. “This is particularly true for technical standards.”
Many businesses criticised the latest draft of the code of conduct for GPAI providers as providing little clarity on the scope of the rules. It is not yet clear, for instance, if the law will also apply to companies fine-tuning a model developed by a GPAI provider.
“It is currently impossible to know exactly what the practical implementation of the AI Act will look like,” says LightOn’s Daudet. “Giving ourselves a bit of time to clarify things is maybe a good thing.”
Peter Sarlin, the CEO of Helsinki-based Silo AI, tells Sifted that “stopping the clock on enforcement by around six months would give companies time to understand and implement the requirements properly.”
Read the orginal article: https://sifted.eu/articles/eu-ai-act-pause-analysis/
