A documented system
Providers establish and document a post-market monitoring system in a manner proportionate to the nature of the AI technologies and the risks of the high-risk AI system.
What Article 72 of the EU AI Act requires from providers once a high-risk AI system is in use, and how it feeds serious incident reporting under Article 73.
Post-market monitoring is the set of activities a provider carries out to collect and review experience from AI systems it has placed on the market or put into service, so it can identify any need for corrective or preventive action.
Article 72 of the EU AI Act requires it for every high-risk AI system.
Providers establish and document a post-market monitoring system in a manner proportionate to the nature of the AI technologies and the risks of the high-risk AI system.
The system actively and systematically collects, documents and analyzes data on performance throughout the system’s lifetime, so the provider can evaluate continuous compliance with the Chapter III, Section 2 requirements.
Where relevant, monitoring includes an analysis of the interaction with other AI systems. Sensitive operational data of law enforcement deployers is excluded.
The system is based on a plan within the technical documentation. As amended in 2026, the Commission is to adopt guidance with a template for the plan by September 2, 2027.
Article 3(49) defines a serious incident as one that leads to death or serious harm to health, serious and irreversible disruption of critical infrastructure, infringement of fundamental rights obligations, or serious harm to property or the environment. Providers report to the market surveillance authority where it occurred.
| Situation | Reporting deadline under Article 73 |
|---|---|
| Any serious incident | Immediately after a causal link, or its reasonable likelihood, is established, and no later than 15 days after becoming aware |
| Widespread infringement, or serious and irreversible disruption of critical infrastructure | Immediately, and no later than two days after becoming aware |
| Death of a person | Immediately after a causal relationship is established or suspected, and no later than 10 days after becoming aware |
Deployers monitor the operation of the high-risk AI system on the basis of the instructions for use and, where relevant, inform the provider in line with Article 72.
Where use may present a risk, the deployer informs the provider or distributor and the market surveillance authority and suspends use. A serious incident goes to the provider immediately.
Deployers keep the logs the system generates automatically, to the extent under their control, for at least six months.
Straight answers, kept current. Last updated .
Ask us anything →The provider of the high-risk AI system. Article 72 places the duty to establish and document the system on providers, and Annex VI makes post-market monitoring one of the things verified during conformity assessment. Deployers supply data and have their own monitoring duty under Article 26(5), but they do not take over the provider’s obligation unless Article 25 makes them the provider.
Not by itself. Article 72 asks for a documented system and plan covering data collection, analysis and continuous compliance, which is an organizational process. Runtime records are one input. Cranium Guardian, for example, inspects traffic between an AI application and the LLM for each interaction and keeps event records a team can use as evidence. No tool is a legal compliance guarantee.
Article 72 applies to providers of high-risk AI systems. As amended in 2026, the classification rules and requirements for those systems apply from December 2, 2027 for Annex III systems and August 2, 2028 for Annex I products, and the Commission’s guidance and template for the monitoring plan are due by September 2, 2027. This page is general information, not legal advice.
Post-market monitoring is done by the provider on its own systems. Market surveillance is done by public authorities, which receive serious incident reports under Article 73 and must take appropriate measures within seven days of receiving one. The two connect: the provider’s monitoring is where reportable incidents surface, and the authority then decides what measures to take.
Primary sources for the figures and frameworks cited on this page. Last verified .
See how Cranium AI discovers, observes, governs, secures and proves every AI model, agent and vendor in your enterprise.
Cookie Policy · Privacy Policy
Global Privacy Control detected. Opt-out request honored: marketing tags stay off in this browser.