Logo gray

General-Purpose AI Code of Practice issued

The General-Purpose AI Code of Practice (‘Code’) was published by the EU Commission and entered into force on 2 August 2025. The essence of the Code is to help companies comply with the provisions of the AI Act. The AI Office will monitor compliance with the rules for new models after one year and for existing models after two years.

The Commission's position is that economic operators who sign this Code demonstrate their compliance with the AI Act. This is expected to reduce administrative burdens and create a more predictable legal environment. The Code is structured in three parts: Transparency, Copyright, and Safety and Security.

As part of the Transparency chapter, a Model Documentation form has been created for companies to use when developing AI. In addition, the Code requires the regular publication of specific information.

The Copyright chapter requires companies to have a copyright policy and to use only information from legal sources when browsing the internet. In addition, AI must take into account copyright restrictions and adequately manage the risk of copyright infringement.

Based on the Safety and Security chapter, AI development and operation must include systemic risk identification and analysis, safety mitigations must be applied, and regular reports must be prepared on the safety of the model used.

The codes of conduct have already been signed by several large companies such as Amazon, Google, IBM, Microsoft, and OpenAI. Opponents of AI regulation and the Code argue that the EU has introduced such strict rules that they make competitiveness impossible.