The European Union on Thursday unveiled a voluntary code of practice for general-purpose artificial intelligence (AI), aiming to help thousands of companies across the 27-member bloc comply with the EU’s upcoming AI Act.
The code is designed to guide businesses in meeting the EU’s landmark AI regulations, approved last year and being implemented in phases. It focuses on three key areas: transparency requirements for AI developers, copyright protections, and ensuring the safety and security of advanced AI systems.
While not legally binding, the code supports the AI Act, whose rules on general-purpose AI—such as OpenAI’s ChatGPT—are set to come into effect on August 2. Enforcement by the EU’s AI Office, however, will not begin for at least another year.
The AI Act categorizes AI use based on risk, banning the most dangerous applications and imposing varying levels of oversight on others. Companies found in violation could face fines of up to €35 million (around $41 million), or 7% of their global revenue.
Some major tech firms, including Meta, have voiced opposition, calling the regulations impractical. U.S. Vice President JD Vance warned earlier this year that overregulation could stifle innovation.
Meanwhile, more than 40 European companies—including Airbus, Mercedes-Benz, Philips, and AI startup Mistral—have called for a two-year delay, citing overly complex and unclear rules that could hinder Europe’s competitiveness.
But Brussels appears determined to move forward. “This is an important step toward making AI in Europe both innovative and safe,” said EU tech chief Henna Virkkunen.
