The European commission’s general-purpose AI Code of Practice: preparing for ai act compliance

On July 10, 2025, the European Commission published a General-Purpose AI (GPAI) Code of Practice for GPAI models. As defined by the AI Act, GAPI models are models that display, “significant generality and are capable of competently performing a wide range of distinct tasks.” The Code is a voluntary framework that was drafted by hundreds of experts over the course of nearly a year. It was created as tool to encourage compliance with the AI Act and does not create any new legal obligations. The Code is broken down into three chapters: (1) Transparency, (2) Copyright, and (3) Safety & Security. While it has been published, it has not been formally approved by the EU.

The Commission will view actions by organizations that sign the Code as good faith efforts and these organizations will not face enforcement actions under the AI Act from August 2, 2025 (the AI Act compliance deadline) through August 1, 2026. Notably, Meta refused to sign the Code. Meta’s Chief Global Affairs Officer recently posted, “This Code introduces a number of legal uncertainties for model developers, as well as measures which go far beyond the scope of the AI Act.” He goes on to say that Meta is not alone; businesses across Europe encouraged the Commission to “Stop the Clock” on implementation of the Code.

The Commission plans to publish guidance in the coming months to provide further clarity on AI Act compliance obligations. Key questions remain regarding how companies should respond, if companies are able to sign portions of the Code, and what enforcement will look like in the coming year.