Source: OJ L, 2024/1689, 12.7.2024

Article 75 Mutual assistance, market surveillance and control of general-purpose AI systems

Summary What does Article 75 of the AI Act say?

This article addresses a specific supervisory challenge that arises at the intersection of general-purpose AI models and the high-risk AI systems built on top of them.

It establishes a division and coordination of oversight responsibilities between the AI Office and national market surveillance authorities, particularly for cases where a single provider develops both the underlying general-purpose AI model and the AI system built upon it.

In that scenario, the AI Office takes on direct supervisory competence.

The article also sets up a mechanism for national authorities to escalate to the AI Office when they hit a wall in their own investigations due to inability to access information about an underlying general-purpose AI model.

Important points:

The AI Office holds direct powers to monitor and supervise compliance where a provider has developed both the general-purpose AI model and the AI system built on it.
Market surveillance authorities are required to cooperate with the AI Office and inform the Board when they have sufficient reason to consider a general-purpose AI system used for high-risk purposes to be non-compliant.
The AI Office is required to supply relevant information to a requesting market surveillance authority within 30 days where that authority cannot complete its investigation due to inability to access information about an underlying general-purpose AI model.

Springlex's summary of the article, a reading aid, not a substitute for the legal text.

1. Where an AI system is based on a general-purpose AI model, and the model and the system are developed by the same provider, the AI Office shall have powers to monitor and supervise compliance of that AI system with obligations under this Regulation. To carry out its monitoring and supervision tasks, the AI Office shall have all the powers of a market surveillance authority provided for in this Section and Regulation (EU) 2019/1020.
1. Where the relevant market surveillance authorities have sufficient reason to consider general-purpose AI systems that can be used directly by deployers for at least one purpose that is classified as high-risk pursuant to this Regulation to be non-compliant with the requirements laid down in this Regulation, they shall cooperate with the AI Office to carry out compliance evaluations, and shall inform the Board and other market surveillance authorities accordingly.
1. Where a market surveillance authority is unable to conclude its investigation of the high-risk AI system because of its inability to access certain information related to the general-purpose AI model despite having made all appropriate efforts to obtain that information, it may submit a reasoned request to the AI Office, by which access to that information shall be enforced. In that case, the AI Office shall supply to the applicant authority without delay, and in any event within 30 days, any information that the AI Office considers to be relevant in order to establish whether a high-risk AI system is non-compliant. Market surveillance authorities shall safeguard the confidentiality of the information that they obtain in accordance with Article 78 of this Regulation. The procedure provided for in Chapter VI of Regulation (EU) 2019/1020 shall apply mutatis mutandis.

Relevant recitals

Recital 156 Powers of market surveillance authorities

In order to ensure an appropriate and effective enforcement of the requirements and obligations set out by this Regulation, which is Union harmonisation legislation, the system of market surveillance and compliance of products established by Regulation (EU) 2019/1020 should apply in its entirety. Market surveillance authorities designated pursuant to this Regulation should have all enforcement powers laid down in this Regulation and in Regulation (EU) 2019/1020 and should exercise their powers and carry out their duties independently, impartially and without bias. Although the majority of AI systems are not subject to specific requirements and obligations under this Regulation, market surveillance authorities may take measures in relation to all AI systems when they present a risk in accordance with this Regulation. Due to the specific nature of Union institutions, agencies and bodies falling within the scope of this Regulation, it is appropriate to designate the European Data Protection Supervisor as a competent market surveillance authority for them. This should be without prejudice to the designation of national competent authorities by the Member States. Market surveillance activities should not affect the ability of the supervised entities to carry out their tasks independently, when such independence is required by Union law.

Recital 161 Responsibility for supervision of general-purpose AI models

It is necessary to clarify the responsibilities and competences at Union and national level as regards AI systems that are built on general-purpose AI models. To avoid overlapping competences, where an AI system is based on a general-purpose AI model and the model and system are provided by the same provider, the supervision should take place at Union level through the AI Office, which should have the powers of a market surveillance authority within the meaning of Regulation (EU) 2019/1020 for this purpose. In all other cases, national market surveillance authorities remain responsible for the supervision of AI systems. However, for general-purpose AI systems that can be used directly by deployers for at least one purpose that is classified as high-risk, market surveillance authorities should cooperate with the AI Office to carry out evaluations of compliance and inform the Board and other market surveillance authorities accordingly. Furthermore, market surveillance authorities should be able to request assistance from the AI Office where the market surveillance authority is unable to conclude an investigation on a high-risk AI system because of its inability to access certain information related to the general-purpose AI model on which the high-risk AI system is built. In such cases, the procedure regarding mutual assistance in cross-border cases in Chapter VI of Regulation (EU) 2019/1020 should apply mutatis mutandis.

Table of contents

Springlex and this text is meant purely as a documentation tool and has no legal effect. No liability is assumed for its content. The authentic version of this act is the one published in the Official Journal of the European Union.