EU Tech Firms Urge Brussels to Delay AI Act Implementation by Two Years

With less than a month remaining before key provisions of the European Union’s Artificial Intelligence Act come into effect, a coalition of both US and European tech firms is urging Brussels to pause the implementation. Groups representing tech giants like Google, Meta, and Mistral are advocating for a delay of several years, and they have garnered support from some politicians.

Over 40 EU tech groups have lobbied for a pause in the implementation of the AI Act. Passed after intense negotiations among member states last year, the AI Act introduces its rules in stages. The next significant milestone is August 2, when obligations for general-purpose AI (GPAI) models will take effect. This category includes foundation models from companies like Google and OpenAI, as well as emerging players like Mistral.

From August 2, developers of GPAI models will be required to compile detailed technical documentation, comply with EU copyright law, and provide summaries of the data used to train their algorithms. Additionally, firms will need to conduct checks for bias, toxicity, and stability before these systems can go live. Models deemed high-impact or posing systemic risk will face even stricter duties, including comprehensive risk assessments, adversarial attack tests, reporting serious incidents to the Commission, and disclosing the energy consumption of their services.

For many AI companies, these new requirements translate into significant compliance costs. Moreover, developers are operating in the dark because the promised guidance to help them meet the new rules has not been provided. The so-called AI Code of Practice, which was supposed to be published by May 2, has not materialized, leaving developers uncertain about how to draft the required paperwork or set up toxicity tests.

In an open letter released last Thursday, 45 European tech groups urged the Commission to impose a two-year “clock-stop” ahead of the August deadline. They argued that rushing ahead without clear standards will only create confusion and potentially stifle innovation in Europe, where regulatory teams are generally more stringent than their US counterparts. The letter also called for the simplification of the rules to prevent startups from being driven overseas by excessive red tape.

So far, the European Commission has not indicated any intention to change the August 2 deadline. However, EU digital chief Henna Virkkunen has pledged to publish the awaited Code of Practice before next month, acknowledging the challenges firms are facing. Political figures have also weighed in, with Sweden’s Prime Minister Ulf Kristersson describing parts of the AI rulebook as “confusing” and calling for a pause. The tech lobby group CCIA Europe similarly advocated for a “stop-the-clock” measure to provide legal certainty until the missing standards are available.

There is precedent for such delays. In 2024, Brussels postponed its Deforestation Regulation by a year, and earlier this spring, it fast-tracked amendments to corporate sustainability directives in under two months. However, to achieve a postponement for GPAI obligations, lawmakers would need to act swiftly. The European Parliament’s last plenary session before the summer recess begins the week of July 7, and any deal must be finalized before the July 10 cutoff. This leaves a narrow window to test whether the EU has both the will and the agility to pause its AI ambitions without derailing them. Tech firms are already concerned that the bloc’s regulations are overly restrictive, with companies like Meta and Google complaining that the rules hinder innovation.