Secondary Legislation / Implementing Acts

The application of Regulation (EU) 2024/1689 is supported by a set of instruments developed by the European Commission and the European AI Office. The aim of these instruments is to operationalise, clarify and support the practical implementation of the obligations set out in the European Artificial Intelligence Regulation (AI Act) throughout the artificial intelligence value chain.

They do not constitute additional legislation. Instead, they are tools designed to support compliance, provide guidance and ensure consistent, harmonised application of the Regulation across the European Union.

Instruments that support the implementation of obligations relating to general-purpose AI systems

The General-Purpose AI Code of Practice. (Published on 10 July 2025)
Guidelines on the scope of the obligations applicable to general-purpose AI (Published on 18 July 2025)
Template for the public summary of training data used for general-purpose AI models. (Published on 24 July 2025)

Transparency and AI-generated content
The Commission has also developed instruments relating to transparency and AI-generated content, including:

Guidelines and a transparency framework under the European Artificial Intelligence Regulation - AI Act. (Subject to consultation until 3 June 2026)
A Code of Good Practice for the marking and identification of AI-generated content, including deepfakes. (Published on 10 June 2026)

Codes of Conduct under Article 95
Article 95 requires the European AI Office and the Member States to encourage and facilitate the development of codes of conduct that promote the voluntary application of requirements to AI systems.

For non-high-risk AI systems, these codes are intended to encourage the adoption of some or all of the requirements applicable to high-risk systems, such as risk management, data governance, and technical documentation.
Voluntary codes may be developed for all AI systems based on clear objectives and key performance indicators. These may include elements such as:
- Union ethical guidelines for trustworthy AI;
- Assessment and minimisation of environmental sustainability impacts, including energy-efficient programming and design;
- Promotion of AI literacy;
- Support for inclusive and diverse design, including diverse development teams and stakeholder participation;
- Prevention of adverse impacts on vulnerable persons or groups, and promotion of gender equality.
The Commission may also develop sector-specific initiatives aimed at reducing technical barriers to the cross-border exchange of data for AI development.

Commission Guidelines under Article 96
The Regulation also provides for the adoption of guidelines on the practical implementation of the Regulation. These guidelines are expected to cover the following in particular:

The definition of an AI system: Practical application of the legal definition. (Published on 6 February 2025)
Transparency obligations: Practical guidance on applying transparency rules to certain AI systems under Article 50. (Subject to public consultation until 3 June 2026)
Requirements for high-risk AI systems: Applying requirements relating to transparency, risk management, data, documentation and responsibilities across the value chain under Articles 8 to 15 and Article 25. (Subject to public consultation until 23 June 2026);
Prohibited practices: Clarifying the prohibited AI practices set out in Article 5.
Substantial modifications: Practical guidance on what constitutes a 'substantial modification' to existing systems;
Interaction with other legislation: Detailed information on the relationship between the AI Act and the Union harmonisation legislation listed in Annex I.
Ensuring consistency in enforcement of relevant Union laws and responsibilities across the AI value chain: Clarifying the obligations applicable to providers, deployers, importers, distributors and other economic operators.
Application of obligations to systems already on the market: Guidance on updates, integrations or modifications to existing systems.
Supervision and harmonised enforcement of the Regulation: Supporting consistent interpretation by competent authorities and supervisory bodies.
Practical application of documentation and record-keeping requirements: Technical documentation, record-keeping and demonstrating conformity.

Further information on the European Commission’s guidelines supporting the implementation of the AI Act can be found here: https://digital-strategy.ec.europa.eu/en/news/supporting-the-implementation-of-the-ai-act-clear-guidelines.

Implementing Acts and Regulations
The AI Act provides for a number of implementing acts, which are designed to operationalise key aspects of the Regulation.

The Commission is required to adopt a set of mandatory implementing regulations addressing critical elements of the framework, including:

Scientific panel: Rules on the establishment and functioning of the scientific panel of independent experts supporting enforcement activities under Article 68.
Regulatory sandboxes: Detailed arrangements for the establishment, development, implementation, operation and supervision of AI regulatory sandboxes under Article 58.
Real-world testing plans: Detailed elements to be included in real-world testing plans for high-risk AI systems under Article 60.
Post-market monitoring plan template: A template for the post-market monitoring plan and the list of elements to be included for high-risk AI systems under Article 72.
Assessments of general-purpose AI models: Detailed conditions and procedures for assessments carried out by the Commission in relation to general-purpose AI models under Article 92.
Fine procedures for general-purpose AI obligations: Detailed arrangements and procedural safeguards for the imposition of fines on providers of general-purpose AI models under Article 101.

Optional implementing acts
The Commission may adopt further implementing acts where it deems it necessary or where specific conditions are met. These include:

Codes of practice for general-purpose AI: Approval of a code of practice and granting general validity across the Union to demonstrate compliance by providers of general-purpose AI models under Article 56.
Common specifications: Adoption of common specifications for high-risk AI system requirements or obligations relating to general-purpose AI models where harmonised standards are unavailable or insufficient under Article 41.
Codes of practice for transparency: Approval of codes of practice concerning the detection and labelling of AI-generated or manipulated content such as deepfakes, or adopting common rules where such codes are inadequate under Article 50.
Common rules for general-purpose AI: Adoption of common rules to implement obligations applicable to general-purpose AI models where a code of practice cannot be finalised or is deemed inadequate under Article 56.
Notification of bodies: Suspending, restricting or withdrawing the designation of a notified body where it no longer meets the applicable requirements and the relevant Member State fails to take appropriate action under Article 37.