The European Commission announced that the long-awaited Code of Practice meant to help companies comply with the EU’s Artificial Intelligence Act might not take effect until the end of 2025. This delay comes amid growing calls from major tech firms like Google and Meta, European AI companies like Mistral and ASML, and some EU governments to postpone implementation due to the absence of finalized guidance. The AI Act is a landmark regulatory framework aimed at ensuring safe and transparent use of artificial intelligence technologies within the EU.
New AI Code of Practice Sets Voluntary Standards and Quality Benchmarks for GPAI Models
Initially set for release on May 2, the Code of Practice for general-purpose AI (GPAI) models—such as OpenAI’s ChatGPT—will now be introduced in the coming days, with companies expected to sign up next month. The Commission anticipates the code will become operational by the end of this year. While signing up is voluntary, companies that opt out will forgo the legal certainty and regulatory support the code offers, potentially putting them at a competitive disadvantage.
EU Faces Pressure Over AI Rules as Code of Practice Delayed Amid Tech Industry PushbackAI advocacy groups, like The Future Society, emphasized the significance of the Code of Practice as a cornerstone of the EU’s AI rulebook. According to Executive Director Nick Moës, the code sets a clear standard for the quality of services provided by GPAI systems, helping to protect downstream users and prevent misleading practices by tech giants. It also mandates evaluation by independent experts and authorities to ensure the reliability of AI technologies offered in the European market.
European Commission Stands Firm on AI Act Amid Big Tech Pressure and Criticism
Despite the lobbying for delays, the European Commission reaffirmed its commitment to enforcing the AI Act’s goals. The spokesperson stressed the importance of establishing consistent, risk-based rules across the EU and maintaining the safety of AI systems. The Commission rejected calls to slow down the rollout, underscoring its intention to preserve the integrity and protective measures of the regulation.
Activist organizations, such as the Corporate Europe Observatory, criticized the influence of large tech companies in the policy process. Researcher Bram Vranken accused Big Tech of employing a familiar strategy—delaying, pausing, and deregulating—to weaken legislation designed to guard against biased and harmful AI systems. While the AI rules for new GPAI models will be legally binding starting August 2, enforcement won’t begin until August 2026, with existing models having until August 2027 to comply.
