The European Commission is considering postponing some parts of the AI Act. According to several sources, this is in response to growing pressure from the technology sector as well as political pressure from Washington. The information was first published by the UK's The Guardian.
The AI Act, which came into force in August 2024, is the world's first comprehensive legal framework to regulate the use of AI across sectors. The aim is to protect citizens from „high-risk“ systems that could threaten health, safety or fundamental rights. However, most of the key obligations have not yet come into force - they are not expected to come into force until August 2026 or a year later, according to the European Commission's official website.
According to an internal document cited by the Financial Times and The Guardian, the Commission is considering a one-year transition period for providers of generative AI, i.e. systems capable of producing text or images. This would give companies that had already launched their products before the regulation came into force an extra year to adapt their processes without disrupting the market.
At the same time, consideration is being given to postponing fines for breaches of the transparency rules until August 2027 to give AI providers and users sufficient time to adapt. In addition, according to specialist site MLex, developers of „high-risk“ systems could be given more flexibility to monitor the performance of their products - i.e. less strict oversight than originally envisaged by the legislation.
Brussels is being pressured not only by the United States, but also by part of the European industry. Donald Trump's administration has threatened to impose tariffs on countries that impose regulations or digital taxes that „discriminate against American technology“. On the European side, dozens of large companies, including Airbus, Lufthansa and Mercedes-Benz Group, have joined in an open letter calling for a two-year delay in the implementation of the AI Act to ensure a reasonable implementation and simplification of the rules, French daily Le Monde reported. According to the signatories, such a move would show that Europe is serious about its competitiveness and innovation agenda.
Another vocal critic is Meta, which this year refused to sign the commissioner's code for general purpose models. According to the company's vice-president Joel Kaplan, „Europe is heading in the wrong direction on AI“ and the adopted code introduces legal uncertainty.
European Commission spokesman Thomas Regnier stressed that the discussion on the postponement did not mean a weakening of the EU's commitment. „The European Union retains the sovereign right to decide on its own legislation,“ he told Reuters. Regnier said Brussels remains in contact with global partners, but no third country will determine how the EU regulates the technology sector.
The European Commission is due to publish its proposals on 19 November 2025. If the postponement is indeed approved, it could have a major impact not only on technology companies but also on the entire innovation ecosystem in Europe. As Reuters notes, it is a balancing act between protecting citizens and remaining competitive with the US and China.
gnews.cz - GH
English 
Czech 
Spanish 
French 
Italian 
Russian 
Chinese 
German