GovCompass
Responsible AI

Art. 52 EU AI Act: the systemic-risk notification procedure

By GovCompass.ai· · Aligned with the consolidated EU AI Act, including the 2026 Omnibus amendments.

Art. 52 sets out the procedure that connects to the systemic-risk classification of Art. 51. A provider must notify the Commission without delay, and within two weeks, when its general-purpose AI model meets or is foreseen to meet the systemic-risk threshold. The provider can argue, with the notification, that its model does not present systemic risk despite crossing the threshold. The Commission maintains and publishes a list of GPAI models with systemic risk.

The notification duty

Art. 51 defines when a model has systemic risksystemic riskEU AI Act category for the most capable general-purpose models (presumed above a training-compute threshold), triggering extra duties: evaluations, adversarial testing, incident reporting, cybersecurity.Open full entry →; Art. 52 sets out what a providerproviderThe actor who develops an AI system (or has it developed) and places it on the market or into service under its own name — carrying manufacturer-style duties: design controls, documentation, conformity.Open full entry → must do about it. When a general-purpose AI modelgeneral-purpose AI modelEU AI Act term for a model displaying significant generality and capable of many distinct tasks, typically integrated into downstream systems; carries its own obligation set, with extra duties for models posing systemic risk.Open full entry → meets the condition for systemic risk, that is, when it reaches or is reasonably foreseen to reach the 10^25 FLOP compute threshold, the provider must notify the Commission. The notification is due without delay and in any event within two weeks of the moment the provider knows or should foresee that the threshold is met. Reaching the threshold and failing to notify is itself a breach that the Commission's enforcement powers can address.

Contesting the presumption

The systemic-risk threshold is a presumption, and Art. 52 builds in the route to rebut it. Together with the notification, a provider may present arguments to demonstrate that, exceptionally, its model does not present systemic risk despite meeting the compute threshold, because it does not in fact have high-impact capabilities matching the most advanced models. The Commission assesses those arguments. If it does not accept them, the model is classified as systemic risk and the Art. 55 obligations apply.

The procedure also runs in the other direction over time. A provider whose model has been classified as systemic risk can request a reassessment based on new, concrete reasons if circumstances change, so the classification is not permanently fixed once made.

The Commission's list

Art. 52 requires the Commission to publish and keep up to date a public list of the general-purpose AI models classified as having systemic risk. This list is the transparencytransparencyOpenness about the fact that AI is used and how it operates in general: disclosures, documentation, notices. Pairs with explainability, which addresses individual outcomes.Open full entry → mechanism for the systemic-risk tier: it makes visible which models are subject to the heavier Art. 55 regime, which matters to downstream providers who integrate these models and need to know the regulatory status of what they build on.

Why it matters

For the small group of providers whose models reach the frontier, Art. 52 is the operational trigger that turns the Art. 51 classification into a concrete duty with a deadline. For everyone else, the published list is the useful artefact: it is how a downstream organisation can confirm whether a foundation modelfoundation modelA model trained on broad data at scale that can be adapted to many downstream tasks; called a general-purpose AI model in EU AI Act terminology.Open full entry → it relies on is a systemic-risk model carrying the full Art. 55 obligations.

In the GovCompass-7

Art. 52 is an accountabilityaccountabilityThe principle that a named human or organization answers for an AI system's outcomes, through ownership, documentation, audit trails and redress — never the system itself.Open full entry → and transparency provision. The notification duty assigns clear responsibility to the model provider, and the published list operationalises transparency at the level of the model market, letting downstream parties see the regulatory status of the models they integrate.

Continue reading

Legal referencesArt. 52

More on Accountability

Art. 10 EU AI Act: data and data governance for high-risk AI

Reference

Art. 10 requires that the training, validation, and testing data for high-risk AI systems meets quality criteria: relevant, sufficiently representative, and as free of errors and complete as possible for the intended purpose. It also requires documented data governance practices covering collection, preparation, bias examination, and gap mitigation, and it permits the limited processing of special-category data where strictly necessary to detect and correct bias, under safeguards.

Art. 12 EU AI Act: record-keeping and logging for high-risk AI

Reference

Art. 12 requires high-risk AI systems to technically allow for the automatic recording of events (logs) over their lifetime. The logging must enable traceability of the system's functioning at a level appropriate to its intended purpose, support post-market monitoring, and help identify situations that may lead to risk or substantial modification. It is a design obligation on the provider that makes the system auditable by construction.

Art. 19 EU AI Act: keeping the automatically generated logs

Reference

Art. 19 requires providers of high-risk AI systems to keep the logs that the system automatically generates (under Art. 12) for as long as they control them, for a period appropriate to the intended purpose and at least six months unless other law requires longer. It is the retention counterpart to the Art. 12 logging capability, and it works alongside the deployer retention duty in Art. 26.6.

Art. 26.1 EU AI Act: following provider instructions as a deployer

Reference

Art. 26.1 requires deployers to use high-risk AI systems strictly in accordance with the provider's instructions for use. This means using the system only for its intended purpose, within its specified technical configuration, and by qualified users, and documenting that compliance. Deviating from the instructions can shift liability entirely to the deployer.

More on Transparency & explainability