The EU AI Act: Authorised Representatives
The EU Artificial Intelligence Act (“the Act”) introduces the concept of AI “authorised representatives” and in this blog, we explain how they are set to play a key role in assisting certain organisations established outside the EU in complying with their obligations under the Act when they offer their AI systems or AI models on the EU market.
Read our blog for an overview of the EU AI Act.
Who must appoint an authorised representative?
The requirement to appoint an authorised representative applies to providers established outside the EU and who have either developed a high-risk AI system or general-purpose AI model or have had one developed for them that they intend to place on the EU market under their own name or trademark.
However, providers of general-purpose AI models that are released under a free and open-source licence that allows access, usage, modification and distribution of the model, and whose parameters and information on model usage are made publicly available, do not have to appoint an authorised representative, unless those AI models present systemic risks.
What is an authorised representative?
An authorised representative is any individual or legal entity located within the EU that has received a written mandate from the type of provider referred to in the paragraph above to carry out, on its behalf, the obligations within the Act.
Essentially, authorised representatives will act as intermediaries between non-EU established AI providers and EU competent authorities on matters relating to compliance with the Act.
The Act does not identify specific qualifications an authorised representative must have in order to act on behalf of a non-EU provider. However, similar to EU representatives under the EU GDPR, authorised representatives under the Act must have a strong understanding of the Act, the obligations under it, and technical knowledge of AI systems.
When do you need an authorised representative under the AI Act?
Authorised representatives must be appointed before the high-risk AI system or AI model is placed on the EU market i.e., before it is launched in the EU and made accessible to EU deployers and users.
What is the role of authorised representatives?
Authorised representatives will perform a key role under the Act and will have several responsibilities to ensure compliance with the Act. There are minor differences between the duties of authorised representatives appointed in relation to high-risk AI systems compared to those appointed in relation to AI models and we, therefore, combine them below:
- They must act on behalf of the provider and fulfil the obligations under the Act.
- They will be responsible for verifying that the EU declaration of conformity and technical documentation is in place and that the provider has conducted an appropriate conformity assessment procedure.
- They must keep, for a period of 10 years, the contact details of the provider, EU declaration of conformity, technical documentation, and relevant certificates. These documents must be made available to the competent authority or AI Office, upon request.
- They must cooperate with competent authorities or the AI Office in any action taken in relation to high-risk AI systems or general-purpose AI models.
- They must ensure high-risk AI systems have been correctly registered on the EU database. The registration may be carried out by the provider themselves or the authorised representative.
- They must be listed as the contact person, solely or in addition to the provider, and liaise with the competent authorities on all matters related to ensuring compliance with the AI Act.
- They must terminate the mandate if they believe the provider is not performing its obligations under the Act. In such cases, they must inform the market surveillance authority and, where applicable, the relevant notified body, of the termination and the reasons for it.
What do you need to do to comply?
The Act is due to come into force by the end of August 2024 and will take effect in stages. It is expected that Article 54, under which the obligations of authorised representatives for general-purpose AI models are set out, will become applicable as early as July next year. Article 22, which relates to authorised representatives for high-risk AI models, is expected to take effect a year later, in July 2026. It is, therefore, important for any non-EU-based organisation seeking to provide their high-risk AI systems and/or AI models in the EU to ensure they obtain specialist advice to understand their requirements in relation to appointing an authorised representative and how to meet their obligations under the Act.
Want to understand your obligations under the EU AI Act?
If you are unclear on what your obligations are under the Act and how to comply, please get in touch, using the form below or visit our AI Governance Services page.
