A new EU Code of Practice for general-purpose AI models, such as ChatGPT and Gemini, is now open for sign-up.
Published by the European Commission on Thursday, the voluntary code is designed to help AI developers align with the AI Act.
The rules on general-purpose AI will begin to apply on 2 August, but it’s still unclear which companies will sign the code.
The document includes three key areas: Transparency, Copyright, and Safety — the last applying only to the most powerful AI models.
By signing the code, companies are expected to align more easily with the AI Act and face fewer legal uncertainties.
Those who opt out could face closer scrutiny from regulators once enforcement begins in August.
While the AI Act came into force in August 2024, it will not be fully enforced until 2027 across the European Union.
Many AI developers say they do not have enough time to meet new requirements and have called for more flexibility.
Although the code was originally due in May, its release was delayed amid criticism from tech firms and copyright holders.
Some say the rules could conflict with EU copyright laws and potentially stifle innovation in the growing AI sector.
Earlier this week, CEOs from over 40 European companies urged the EU to delay parts of the AI Act by two years.
The European Commission may consider delaying some rules on high-risk AI systems, but no decision has been made yet.
With the deadline fast approaching, the code’s legal impact and level of industry participation remain highly uncertain.
