Delayed Guidance for Companies Facing New EU AI Rules
A critical
Code of Practice intended to assist thousands of companies in complying with the European Union's stringent AI regulations may not be ready until the
end of 2025, according to recent updates from the European Commission[2][5]. The Code is designed to offer much-needed guidance as the landmark AI Act's requirements move closer to enforcement.
Industry Lobbying and Implementation Concerns
Major technology firms—including
OpenAI’s ChatGPT, Google, Meta, as well as European companies such as Mistral—alongside several EU governments, have been pushing for a delay in the implementation of the AI Act. The primary reason cited is the absence of this vital Code of Practice, which was initially scheduled for publication on May 2, 2025[2][5].
A spokesperson for the European Commission revealed that while a presentation of the code is imminent, the expected timing for companies to formally sign up may be next month, with the guidance likely to take effect by the end of the year. The European AI Board is actively discussing this timeline[2][5].
What Is Covered by the Code of Practice?
The Code of Practice specifically targets
General Purpose AI (GPAI) models, which include large language models powering systems like
ChatGPT and Google’s AI solutions.
-
Transparency and Copyright Requirements: All GPAI providers must adhere to strict rules on transparency and copyright compliance[1][3].
-
Safety and Security for High-Risk Models: Additional obligations apply for providers of GPAI models that could pose systemic risks[1].
A multi-stakeholder process—featuring independent experts, AI providers, civil society, and government representatives—has shaped the Code's content to cover comprehensive compliance pathways[1][3].
Voluntary Participation Offers Legal Certainty
Enrolling in the Code of Practice is voluntary. However, companies opting out may forgo certain legal certainties provided to signatories, exposing themselves to greater compliance risk under the new regulatory framework[2][5]. Some Big Tech firms have already indicated plans not to participate, seeking alternative avenues to meet the AI Act’s demands.
Broader Implications for AI in Europe
The AI Act’s core regulations for general-purpose AI models are slated to become effective in August 2025[3]. The Code of Practice is anticipated to serve as a central tool, helping organizations demonstrate compliance and remain at the forefront of state-of-the-art practices.
AI advocacy groups, such as The Future Society, have noted that the Code is set to become a critical part of the EU's wider AI rulebook, ensuring clarity and consistency for businesses operating with advanced AI tools[2].
