Service Overview
Are you placing a general-purpose AI model on the EU market from outside the Union?
If your organisation provides a general-purpose AI model — or a general-purpose AI model with systemic risk — and is established outside the European Union, Article 54 of the EU AI Act requires you to designate, by written mandate, an authorised representative established in the Union before that model is placed on the EU market. The representative is the formal point of contact for the European AI Office and national competent authorities throughout the model's market lifecycle.
A general-purpose AI model means an AI model that displays significant generality and is capable of competently performing a wide range of distinct tasks, regardless of how it is placed on the market and which can be integrated into a variety of downstream systems or applications. Where the model meets the criteria in Article 51 — including the cumulative compute threshold for training — it is classified as a general-purpose AI model with systemic risk and additional obligations apply under Article 55.
Service Detail
What this service covers
Our focus
A serious EU foothold for foundation model providers
Lionheart provides Article 54 representation through our Dublin-based EU entity. We act as a structured compliance function for non-EU providers of general-purpose AI models — receiving and escalating correspondence from the European AI Office, holding documentation as required, and supporting cooperation with authorities on the model's compliance posture.
Built for the documentation duties Article 54 actually imposes
The Article 54 mandate is precise. The representative must verify that the technical documentation specified in Annex XI has been drawn up and that all obligations under Article 53 have been complied with by the provider; keep a copy of the technical documentation and the contact details of the provider available to the AI Office and national competent authorities for ten years after the model has been placed on the market, together with the mandate; cooperate with the AI Office and competent authorities on any action they take in relation to the model; and where applicable, comply with the obligations relating to the registration of providers in Article 49 paragraph (1).
Additional duties for systemic-risk models
Where the model is classified as a general-purpose AI model with systemic risk, additional obligations under Article 55 apply to the provider — including model evaluation, systemic risk assessment and mitigation, serious incident reporting to the AI Office, and an adequate level of cybersecurity protection. The authorised representative remains the formal interface for these communications. Lionheart's procedures are designed to handle the heightened reporting cadence these models require, including the serious-incident escalation path.
Mandate termination where the provider acts unlawfully
Article 54(3) imposes the same termination duty found in Article 22: where the representative has reason to consider that the provider is acting contrary to its obligations under the Act, it must terminate the mandate and immediately inform the AI Office. We treat this as a substantive part of the role and explain our internal procedures before any mandate is signed.
Open-source carve-out — read carefully
The AI Act provides an exemption for general-purpose AI models released under a free and open-source licence, where the parameters, including the weights, the information on the model architecture and the information on model usage are made publicly available, allowing access, usage, modification and distribution of the model. The exemption does not apply to general-purpose AI models with systemic risk. Whether a particular release qualifies for the exemption is a fact-specific assessment; Lionheart can help you map the analysis.
Our services include
Formal designation
Acting as your EU Authorised Representative under Article 54 of the AI Act, by written mandate, with clearly documented scope and procedures.
Ireland-based contact details
Lionheart's Dublin address and a dedicated representative email address, available for inclusion in your model's documentation, the information you make available to downstream providers under Article 53, and any required public disclosures.
Annex XI documentation verification
Verifying that the technical documentation specified in Annex XI has been drawn up and that the obligations under Article 53 have been met before the model is placed on the EU market.
10-year documentation safekeeping
Holding a copy of the technical documentation, your contact details as provider, and the mandate, available to the AI Office and national competent authorities for ten years after the model has been placed on the market.
AI Office liaison
Acting as the formal contact point for the European AI Office and national competent authorities — receiving, logging and forwarding all correspondence without delay.
Systemic-risk reporting
For models designated as general-purpose AI with systemic risk: structured escalation procedures for serious incidents reported to the AI Office under Article 55, and supporting communications on model evaluation, risk assessment and mitigation.
Article 53 cooperation
Supporting your cooperation with the AI Office and competent authorities on requests linked to your Article 53 obligations — copyright policy, training-data summary, downstream provider information.
Escalation procedures
Defined escalation paths and service levels for time-sensitive regulatory communications and information requests.
Coordinated mandates
Single-relationship handling where Article 54 obligations overlap other Lionheart EU representative services — DSA, GDPR, NIS2 — under one engagement and coordinated escalation path.
Legal basis: Article 54, EU Artificial Intelligence Act (Regulation (EU) 2024/1689). General-purpose AI provider obligations are set out in Article 53; classification of general-purpose AI models with systemic risk is governed by Articles 51 and 52; additional obligations for systemic-risk providers are set out in Article 55. Annex XI specifies the technical documentation required.
Exemptions:
Article 54 does not apply to general-purpose AI models released under a free and open-source licence as defined in the Act, except where the model is classified as having systemic risk. AI used solely for scientific research and development before placing on the market is also outside scope. Whether your release qualifies turns on the specific licence terms and what is published; Lionheart can provide a preliminary assessment.
Get started
Not sure if you are in scope?
Use the guided self-check tool to map which representative obligations may apply.