With less than a month remaining before key parts of the European Union’s AI Act take effect, several major tech companies are urging the European Commission to delay the implementation. U.S. giants such as Google parent Alphabet and Meta, along with European firms like Mistral and ASML, are among those pushing for a pause in the upcoming rules.
The provisions impacting general purpose AI (GPAI) models are scheduled to come into force on August 2, according to a Commission spokesperson. However, enforcement powers will not apply until August 2, 2026. The AI Act, passed in 2024 following extensive negotiations among EU member states, includes staggered implementation of its various components.
The upcoming deadline includes new transparency obligations for foundation models, such as those developed by OpenAI, Google, and Mistral. These rules require companies to provide technical documentation, summaries of training data content, and proof of compliance with EU copyright laws. Developers must also test their AI systems for bias, toxicity, and robustness prior to release.
For high-impact GPAI systems and those considered to pose systemic risks, the requirements are even stricter. Companies will be expected to conduct model evaluations, perform adversarial testing, report serious incidents to the Commission, and disclose information about energy usage.
However, many companies say they are unprepared to meet the obligations due to a lack of clear guidance. The long-anticipated AI Code of Practice—meant to help developers align with the AI Act—missed its expected release on May 2. In response, a coalition of 45 European companies published an open letter, stating: “To address the uncertainty this situation is creating, we urge the Commission to propose a two-year ‘clock-stop’ on the AI Act before key obligations enter into force.”
There are also concerns the legislation could hinder innovation, particularly for smaller European firms that lack the resources of their U.S. counterparts. Some political leaders, including Swedish Prime Minister Ulf Kristersson, have echoed these concerns and called for a pause.
The Commission has indicated that guidance, including the AI Code of Practice, may now be delayed until the end of 2025. “A bold ‘stop-the-clock’ intervention is urgently needed to give AI developers and deployers legal certainty, as long as necessary standards remain unavailable or delayed,” said industry lobbying group CCIA Europe.
