Will the EU delay enforcing its AI Act?

As the August 2 deadline approaches for parts of the European Union's AI Act, a growing chorus of companies and politicians is advocating for a postponement of its enforcement. Major players in the tech industry, including U.S. giants like Google and Meta, alongside European firms such as Mistral and ASML, have petitioned the European Commission for a significant delay in the act's implementation. The AI Act, which was passed after extensive negotiations among EU members a year ago, is designed to introduce regulations in phases over several years. The initial provisions, particularly concerning general purpose AI (GPAI) models, are set to come into effect next month. These regulations will require companies to adhere to transparency standards, including the preparation of technical documentation, compliance with EU copyright laws, and providing comprehensive summaries of the data used for training their algorithms. In addition, companies must evaluate their AI models for bias, toxicity, and robustness prior to their launch. Those classified as high-risk will have to undertake model evaluations, risk assessments, and adversarial testing, while also reporting serious incidents to the European Commission and sharing details regarding their energy consumption. The push for a delay stems from concerns about the financial burden of compliance, especially for AI model creators facing stringent requirements. The absence of guiding documentation further complicates matters, as the AI Code of Practice—a crucial resource for developers—has yet to be published, missing its initial May 2 release date. A coalition of 45 European companies expressed their concerns in an open letter, urging the Commission to implement a two-year 'clock-stop' on the act to facilitate compliance. They also emphasized the need for a simplification of new regulations to prevent hindering innovation, particularly in Europe where tech firms often have smaller compliance teams relative to their U.S. counterparts. While the European Commission has not yet provided any indication of a delay, EU tech chief Henna Virkkunen has committed to releasing the AI Code of Practice before the start of next month. Political figures, including Swedish Prime Minister Ulf Kristersson, have echoed the call for a pause, labeling the current regulations as perplexing. The tech advocacy group CCIA Europe has also stressed the urgent need for a 'stop-the-clock' intervention to ensure legal clarity for AI developers until standard guidelines are made available. At this time, the European Commission has not responded to requests for further comment on the potential postponement of the AI Act's enforcement.