Publications

AI Act: a look back at the controversial publication of the first general-purpose AI code of practice

On July 10, 2025, the European Commission published the final version of the first general-purpose AI code of practice.

This publication divides the main players in this market: while OpenAi [provider of ChatGPT] has hinted that it will adhere to it, this is not the case for Meta, which announced on July 18, 2025, via its Director of International Affairs, its categorical refusal to adopt it[1].

In this context, UGGC Avocats proposes to present, in five questions and answers, the main measures proposed by the European Commission to help suppliers of general-purpose AI models comply with certain obligations of Regulation (EU) 2024/1689 of the European Parliament and of the Council of June 13, 2024 laying down harmonized rules on artificial intelligence (hereinafter the “Artificial Intelligence Regulation” or the “AI Act”)[2], which will come into force on August 2.

What is a general-purpose AI model?

A general-purpose AI model[1] refers to an AI model, including when trained with a large amount of data using large-scale self-supervision, with three characteristics, namely:

  • it presents a significant generality;
  • it is capable of performing a wide range of distinct tasks, regardless of how it is brought to market,
  • it can be integrated into a variety of downstream systems or applications[2].

What are the obligations covered by the code of good practice, and which operators are responsible for them?

The code of practice covers :

  • the obligation of transparency[1] and respect for copyright[2] incumbent on all suppliers of general-purpose AI models, i.e. any natural or legal person, public authority, agency or any other organization that develops or has developed a general-purpose AI model and places it on the market or puts the AI system into service under its own name or brand, whether in return for payment or free of charge[3];
  • the safety and security obligation[4], which only applies to suppliers of general-purpose AI models that present a systemic risk[5].

What are the main measures proposed by the code of good practice to achieve compliance?

ObligationMain measures proposed
TransparencySet up and keep up to date the technical documentation relating to the AI model concerned[1];
Ensure the transmission of relevant information;
Guarantee the quality, integrity and security of information.

[1] A model of technical documentation is provided in the appendix to the general-purpose AI code of practice, accessible via the following link: https://www.uggc.com/app/uploads/sites/2/2025/07/code-of-practice-for-generalpurpose-ai-models-transparency-chapter-2-ujmhh8ifqfhcdp8z5bf9slpxddc-118120.pdf
Respect for copyrightImplement and keep up to date a policy aimed at complying with European Union law on copyright and related rights;
Reproduce and extract only content protected by copyright and related rights but free to use;
Identify and respect rights reservations made by copyright and related rights holders;
Prevent the risk of generating content that infringes copyright and neighboring rights;
Define a point of contact and set up a complaints service in the event of infringement of copyright and neighboring rights.
Safety and securityImplement and update a safety and security framework;
Identify systemic risks;
Define a scenario of possible systemic risks.

Is adherence to the code of good practice compulsory?

No, this is done on a voluntary basis, which is regrettable in a way.

What’s the point of subscribing to the code of good practice?

Adherence to the Code of Practice should enable suppliers to benefit from a presumption of compliance with the IA Act obligations covered by the Code.

Suppliers who choose not to join will be able to prove their compliance by other means, but the European Commission has already announced that they will be subject to stricter controls.

Our firm and its team of specialized lawyers are naturally at your disposal to help you comply with the IA Act.

Sources

[1] Posted on July 18, 2025 on the Linkedin account of Joel Kaplan, Director of International Affairs at Meta :

Europe is heading down the wrong path on AI. We have carefully reviewed the European Commission’s Code of Practice for general-purpose AI (GPAI) models and Meta won’t be signing it. 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. Businesses and policymakers across Europe have spoken out against this regulation. Earlier this month, over 40 of Europe’s largest businesses signed a letter calling for the Commission to ‘Stop the Clock’ in its implementation. We share concerns raised by these businesses that this over-reach will throttle the development and deployment of frontier AI models in Europe, and stunt European companies looking to build businesses on top of them.

Free translation: ” Europe is heading down the wrong path when it comes to artificial intelligence. We have carefully examined the European Commission’s Code of Conduct on General Purpose AI Models (GPAI), and Meta will not be signing it. The Code introduces many legal uncertainties for model developers, as well as measures that go far beyond the scope of the AI Act. Companies and policymakers across Europe have opposed the regulation. Earlier this month, more than 40 of Europe’s largest companies signed a letter calling on the Commission topause its implementation.We share theconcernsexpressedbythese companies that thisregulatoryoverreachrisksstifling thedevelopment anddeployment of themostadvancedAImodelsin Europe, and holding backEuropeancompanieswishing to build businesses around these technologies.”

https://www.linkedin.com/posts/joel-kaplan-63905618_europe-is-heading-down-the-wrong-path-on-activity-7351928745668055042-XuF7?utm_source=li_share&utm_content=feedcontent&utm_medium=g_dt_web&utm_campaign=copy.

[2] Article 3-3 of the IA Act.

[3] Section 3-63 of the IA Act. [1] Excluding AI models used for research, development or prototyping prior to market launch.

[4] Excluding AI models used for research, development or prototyping prior to market launch.

[5] Sections 53(1)(a), 53(1)(b), 53(2), 53(7) of the IA Act.

[6] Article 53-1 c) of the IA Act.

[7] Article 3-3 of the IA Act.

[8] Articles 55(1) and 56(5) of the IA Act.

[9] Section 3-63 of the IA Act.

[10] A model technical documentation is provided as an appendix to the general-purpose AI code of practice, accessible via the following link: https://www.uggc.com/app/uploads/sites/2/2025/07/code-of-practice-for-generalpurpose-ai-models-transparency-chapter-2-ujmhh8ifqfhcdp8z5bf9slpxddc-118120.pdf