The commission’s tech boss Henna Virkkunen said: “in making the most advanced AI models available in Europe innovative, safe and transparent.” 

The code requires AI providers to install “technical measures that prevent their models from generating content that reproduces copyrighted content” and to perform risk testing under the AI act. 

However Big US tech companies have been lobbying hard against these rules, calling them “a disproportionate burden on AI providers” and warning that signatories could be disadvantaged. 

However, large EU firms like Airbus, BNP Paribas and Mistral have urged a two‑year pause, claiming unclear and overlapping laws threaten the bloc’s competitiveness. 

Although voluntary, companies that refuse to sign up may miss out on legal certainty, experts warn. 

The code still needs a thumbs‑up from the European Commission and members of EU states, and enforcement timelines vary. The AI act came into force in August 2024 but many provisions take effect in future waves. For example, general‑purpose AI obligations kick in on 2 August 2025, with a full rollout through 2026 and 2027. 

Commission officials are looking to simplify the staggered approach and streamline implementation.

The move shows the EU is playing tough in the AI regulation game as launch day looms next month.