European Commission publishes GPAI Code of Practice
Twenty-one companies signed the voluntary code covering transparency, copyright and safety; xAI signed only the safety chapter and Meta announced days later it would not sign at all.
- Government & policy
- Notable
The European Commission published the General-Purpose AI Code of Practice, a voluntary compliance tool drafted by independent experts through a multi-stakeholder process, meant to give providers of general-purpose AI models a defined path to demonstrate compliance with obligations under the EU AI Act. It covers three areas: transparency, including a standard Model Documentation Form; copyright, offering practical measures for complying with EU copyright law; and safety and security, setting out risk-management practices for the most advanced models.
The Commission said 21 companies signed the full code, including Amazon, Google, Microsoft, OpenAI, Anthropic and Mistral AI; xAI signed only the safety and security chapter. Both the Commission and the EU AI Board endorsed the code as an adequate route to compliance, which becomes mandatory for the largest “systemic risk” models by August 2027. Signing remains voluntary — providers can instead attempt to demonstrate compliance through other means, an option the Commission’s own framing suggested would be more burdensome.
The code’s authority as an industry standard was undercut a week later when Meta announced it would not sign. Chief global affairs officer Joel Kaplan said the Commission’s document introduced “legal uncertainties for model developers” and went “far beyond the scope of the AI Act,” calling it regulatory overreach that would stunt company growth. The split — most major labs signing, Meta refusing outright and xAI signing only partially — left the code operating as a norm most of the industry accepted rather than a rule all of it followed, a gap the same week’s Anthropic transparency proposal was explicitly trying to close through binding US legislation instead.