PARIS (Reuters) – The European Union has gone too far with artificial intelligence regulations, making it harder for global companies to deploy the technology in the region, said Aiman Ezzat, chief executive of French IT consulting group Capgemini, in an interview.
Ezzat’s comments, ahead of the AI Action summit in Paris on Monday, come amid growing frustration with EU regulations from private players in AI and after U.S. President Donald Trump revoked a 2023 executive order signed by Joe Biden that sought to reduce the risks that artificial intelligence poses to consumers, workers and national security.
The EU says its AI Act is the world’s most comprehensive law governing the use of AI, but it has been criticised by some companies for stifling innovation.
“In Europe, we went too far and too fast on AI regulation,” Ezzat told Reuters, adding that the absence of global standards on regulating AI was “nightmarish”.
“It’s complex for us because we have to look at regulation in every country where we operate, what we can do, what we cannot do, and what’s our responsibility as a developer,” said Ezzat.
One of Europe’s biggest IT services companies, Capgemini has partnerships with firms such as Microsoft, Google Cloud and Amazon Web Services (AWS). Clients include Heathrow Airport and Deutsche Telekom.
Frameworks for AI policy are expected to be among topics under discussion by heads of state and executives from the world’s leading technology firms at the Paris summit.
Ezzat expects regulators and companies meeting at the conference to try to come to some alignment on policy, he said. While the EU’s AI Act will take years to be fully implemented, European data protection authorities are already concerned that some AI actors could breach privacy laws.
Recently, the data protection bodies of half a dozen European countries issued requests for information or started analysis on DeepSeek. The Chinese startup shocked industry and global markets last month by showing it could compete with U.S. heavyweights in the field for a fraction of the cost. “It’s a game changer, because it brings something new … it’s more open, so you can do more with it and fine tune it much better,” Ezzat said of the Chinese firm. However, he added that even though DeepSeek shares its models via open source, the transparency was limited.
“Is DeepSeek completely open source? No, absolutely not,” he said, with no access to the datasets used to train the models, for example.
Capgemini has begun discussing implementing DeepSeek’s models with clients, said Ezzat, but noted it was in the “very early stages”.
(Reporting by Florence Loeve and Supantha Mukherjee; Editing by Dominique Patton and David Holmes)
