Violating AI laws can result in 7% of a company's annual revenue, or 3% of companies developing advanced AI models, being fined.
Published Thu, July 10, 2025, 08:02 PM
[BRUSSELS] The European Union has published codes of practice that will help businesses adhere to its groundbreaking AI laws, including copyright protection for creators and advanced model transparency requirements.
The code requires developers to provide up-to-date documentation explaining AI capabilities to regulators and third parties seeking to integrate them into their own products, the European Commission said on Thursday (July 10). Companies will also be banned from training AI on pirated materials, and must respect requests from writers and artists to keep copyrighted work out of their datasets. When AI generates material that infringes copyright rules, the code of practice must institute a process for businesses to address it.
The code of practice is voluntary and aims to help businesses establish internal mechanisms for implementing AI laws. The regulations in effect on amazing timetables establish AI curbs in general purpose and high risk fields and limit several applications. Rules that affect “general AI” such as Openai's ChatGpt and Anthropic's Claude will be applied from next month.
Violating AI laws can result in 7% of a company's annual revenue, or 3% of companies developing advanced AI models, being fined.
The code, which still requires a final sign-off from the Commission and EU member states, is controversial and has sparked backlash from some technology companies, including the metaplatform and the alphabet. This month, European companies, including ASML Holdings, Airbus and Mistraral AI, called for two years to suspend the implementation of the AI Act this month, calling for a more “innovation-friendly regulatory approach.”
The committee has missed an initial deadline for publishing the Code of Practice, but has so far refused to postpone implementation. The code was drafted under the guidance of officials of the Committee, the EU's administrative division that formed a working group consisting of representatives from AI labs, technology companies, academia and digital rights organizations.
The committee will only begin direct supervision of AI Act applications in August 2026. Enforcement is left in the hands of a national court with the specific technical expertise up until then. The committee said that when it signed the Code of Practice, businesses would “enhance legal certainty.” Bloomberg
Share your feedback about BT products and services
