Trade War

EU rules follow general-purpose AI models supplied through APIs and cloud services

Overseas model providers face documentation and representation duties, while a voluntary compliance code and separate output rules shape access to the EU market.

The law 17 November 2026

Europe's general-purpose AI rules reach models supplied remotely, making documentation and legal representation conditions of cross-border technology business. The Commission's guidance says models can enter the EU market through an application programming interface, a download or a cloud service. A provider does not need to ship a physical product to fall within the framework.2

The Commission's code-of-practice guidance distinguishes the start of model obligations in August 2025 from enforcement with fines beginning on 2 August 2026. Models placed on the market before 2 August 2025 have a separate compliance deadline of 2 August 2027. These dates concern general-purpose models, rather than every AI application built from them.1

A voluntary route for demonstrating compliance

The endorsed General-Purpose AI Code of Practice provides a voluntary way to demonstrate compliance. Its transparency chapter includes a model-documentation form, while its copyright chapter addresses the provider's copyright policy. The safety and security chapter concerns the narrower group of models with systemic risk. The Commission's current page records that some providers signed the whole code while xAI signed only its safety and security chapter, leaving the other obligations to be demonstrated through alternative means.5

Covered model providers must prepare technical documentation, supply information to businesses incorporating the model, maintain a copyright-compliance policy and publish a training-content summary. Non-EU providers generally need an authorised representative in the Union before placing the model on its market. Certain open-source models qualify for limited exemptions, but those exemptions do not apply to systemic-risk models.2

Transparency concerns the delivered service

Separate Commission guidance explains that transparency duties can reach overseas AI-system providers where their output is used in the EU. Systems that interact directly with people must disclose that interaction with AI unless it is obvious. Generative systems also face requirements for machine-readable identification of synthetic content.3

The guidance identifies exclusions or qualifications for source code, exclusively machine-to-machine outputs and some closed industrial environments. It also separates the provider designing a system from the business deploying it. Model documentation and a visible notice to an end user address different points in the supply chain.3

Data protection remains alongside the AI rules

France's data regulator, CNIL, says the AI Act complements the General Data Protection Regulation. Its explanation states that the AI legislation:

ne remplace pas les exigences du RGPD.

does not replace the GDPR's requirements.4

Personal-data processing during both model development and deployment can remain subject to data-protection obligations. Providing model documentation consequently does not establish that training data or a deployed service complies with the separate privacy framework.4

The code therefore gives suppliers a route for explaining their model's compliance to regulators and downstream customers. It does not add new legal obligations or replace the binding legislation, according to the Commission's explanation of its status.1

Sources

  1. European Commission. Questions and answers on the code of practice for General-Purpose AI. Updated 20 July 2026European Union's camp · government guidance
  2. European Commission. Guidelines on obligations for General-Purpose AI providers. Accessed 17 November 2026European Union's camp · government guidance
  3. European Commission. Transparency obligations under Article 50 of the AI Act. Accessed 17 November 2026European Union's camp · government guidance
  4. CNIL. Entrée en vigueur du règlement européen sur l'IA : les premières questions-réponses de la CNIL [Entry into force of the European AI regulation: CNIL's first questions and answers]. Updated 17 August 2026France's camp · data protection regulator · in French
  5. European Commission. The General-Purpose AI Code of Practice. Updated 7 October 2026European Union's camp · government implementation information

All articles