Appearance
Post-market monitoring and serious incidents
Conformity assessment is the gate into the EU market; Article 72 post-market monitoring and Article 73 serious-incident reporting are how the regime stays current once the system is in operation. The two Articles work as a pair: Article 72 establishes the systematic data-collection and analysis duty across the lifecycle; Article 73 is the specific reporting obligation triggered when an incident clears the serious incident threshold under Article 3(49).
Article 73 EU AI Act is not Article 33 GDPR. They cover different events, run on different deadlines, and report to different authorities. A single incident can trigger both duties — they do not substitute for one another.
Quick decision
- You are a provider of a high-risk AI system → establish the Article 72 post-market monitoring plan as part of the technical documentation (Annex IV point 9). Set up the data flow from deployers and from your own observability to support continuous Articles 8–15 compliance evaluation.
- You are a deployer and you observe a malfunction → Article 26(5) — inform the provider or distributor and the relevant market-surveillance authority without undue delay; suspend use if the system presents a risk under Article 79(1).
- A death occurred → Article 73(4) — report immediately after establishing or suspecting a causal link, and not later than 10 days after becoming aware.
- Widespread infringement or serious + irreversible disruption of critical infrastructure (Article 3(49)(b)) → Article 73(3) — report immediately and not later than 2 days after becoming aware.
- Other serious incident (Article 3(49)(a)–(d)) → Article 73(2) — report immediately after establishing a causal link or reasonable likelihood of such a link, and not later than 15 days after becoming aware.
- The personal data of natural persons was involved → Article 33 GDPR runs in parallel, on the controller, to the supervisory authority, within 72 hours. The Article 73 duty is separate and not satisfied by the GDPR notification.
TL;DR
- Article 72 — providers establish and document a post-market monitoring system proportionate to the system's risks, based on a post-market monitoring plan (Annex IV point 9). The Commission template is to be adopted by 2 February 2026.
- Article 73 — providers report serious incidents as defined in Article 3(49) to the market-surveillance authority. Each deadline runs immediately + outer cap: immediately + 15 days (default); immediately + 10 days (death); immediately + 2 days (widespread infringement or serious + irreversible disruption of critical infrastructure).
- Article 26(5) / (6) — deployers monitor operation per the instructions for use under Article 26(5); for serious incidents they inform first the provider, then the importer or distributor and the relevant market-surveillance authority. If the provider cannot be reached, Article 73 applies mutatis mutandis. Article 26(6) requires log retention for at least six months.
- Article 73 ≠ Article 33 GDPR — different events, different deadlines, different recipients, different addressees.
- Article 55(1)(c) — for GPAI models with systemic risk, the serious-incident report goes to the AI Office, not the market-surveillance authority. Distinct from the system-level Article 73 duty.
AI Omnibus provisional agreement (May 2026)
Status: provisional political agreement, pending formal adoption
On 7 May 2026 the Council presidency and European Parliament negotiators reached a provisional political agreement on the Digital Omnibus on AI (originally proposed by the Commission on 19 November 2025, part of the 'Omnibus VII' simplification package). The deal must still be formally endorsed by the Council and the Parliament and undergo legal/linguistic revision before adoption. This framework page will be updated once the Omnibus is formally adopted. Until then, the existing EU AI Act text remains legally binding.
Relevant to this page (all conditional on formal adoption):
- Annex III post-market-monitoring and serious-incident-reporting duties would shift to 2 December 2027 (from 2 August 2026); Annex I would shift to 2 August 2028 (from 2 August 2027). The substantive Article 72 / 73 duties are unchanged.
- AI Office competence over GPAI-system supervision would be clarified — where the GPAI model and the AI system come from the same provider, the AI Office would supervise the system, with carve-outs (law enforcement, border management, judicial authorities and financial institutions) where national authorities would remain competent. The Article 73 serious-incident report on the high-risk AI system would continue to go to the market-surveillance authority; the Article 55(1)(c) report on the systemic-risk GPAI model would continue to go to the AI Office.
Until adopted, the published 2024/1689 duties, deadlines and recipients remain the binding reference.
Primary source
Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 — EUR-Lex CELEX 32024R1689 · OJ L, 12.7.2024 · Articles 3(49), 26(5)–(6), 72, 73, 79; Annex IV point 9. AI Omnibus provisional-agreement statements draw on the Council press release of 7 May 2026; not legally binding until adoption.
Article 3(49) — what counts as a 'serious incident'
'serious incident' means an incident or malfunctioning of an AI system that directly or indirectly leads to any of the following:
(a) the death of a person, or serious harm to a person's health;
(b) a serious and irreversible disruption of the management or operation of critical infrastructure;
(c) the infringement of obligations under Union law intended to protect fundamental rights;
(d) serious harm to property or the environment;
— Article 3(49), Regulation (EU) 2024/1689
The trigger is the impact, not the technology. Whether the AI system was the proximate cause or contributed indirectly, the four impact categories drive the Article 73 reporting clock. A near-miss is not a serious incident under Article 3(49); a malfunction without one of the four impacts is not reportable under Article 73 (though it may still be relevant for Article 72 post-market-monitoring analysis).
Article 72 — post-market monitoring system
Providers shall establish and document a post-market monitoring system in a manner that is proportionate to the nature of the AI technologies and the risks of the high-risk AI system.
The post-market monitoring system shall actively and systematically collect, document and analyse relevant data which may be provided by deployers or which may be collected through other sources on the performance of high-risk AI systems throughout their lifetime, and which allow the provider to evaluate the continuous compliance of AI systems with the requirements set out in Chapter III, Section 2. Where relevant, the post-market monitoring shall include an analysis of the interaction with other AI systems.
— Article 72(1)–(2), Regulation (EU) 2024/1689
Article 72(3) — the post-market monitoring plan
The post-market monitoring system shall be based on a post-market monitoring plan. The post-market monitoring plan shall be part of the technical documentation referred to in Annex IV. The Commission shall adopt an implementing act laying down detailed provisions establishing a template for the post-market monitoring plan and the list of elements to be included in the plan by 2 February 2026.
— Article 72(3), Regulation (EU) 2024/1689
Annex IV point 9 requires the technical documentation to include a detailed description of the post-market monitoring system in place, including the post-market monitoring plan. The detailed elements of the plan — fields, indicators, methodology — are to be set by the Commission's Article 72(3) implementing-act template (deadline 2 February 2026). Until the implementing act lands, the provider authors the plan to satisfy the Article 72(1)–(2) substantive duties.
Article 72(4) — sectoral integration and reuse
Article 72 has four paragraphs. Article 72(4) governs sectoral integration in two sub-paragraphs:
- First sub-paragraph — for high-risk AI systems covered by Union harmonisation legislation in Annex I Section A, where a post-market monitoring system and plan are already established under that legislation, the provider may integrate the AI Act post-market monitoring elements into the existing system and plan rather than duplicating them.
- Second sub-paragraph — for high-risk AI systems referred to in Annex III point 5 placed on the market or put into service by financial institutions subject to Union financial-services internal-governance / process rules, the post-market monitoring system is considered to be in place if it satisfies those sectoral internal-governance arrangements.
What "continuous" means in practice
The Regulation does not impose real-time monitoring of every metric. It imposes a system — defined processes, proportionate to risk — that:
- Collects performance and incident data from the deployment context, deployer feedback, and the provider's own observability.
- Analyses the data against the Articles 8–15 requirements and against the intended-purpose and state-of-the-art anchors.
- Feeds back into the Article 9 risk-management system (Article 9(2)(c) makes the post-market data hook explicit) and into the Article 11 / Annex IV technical documentation keep-up-to-date duty.
- Triggers the Article 20 corrective-actions duty when a non-conformity is detected: bring into conformity, withdraw, disable, or recall as appropriate.
Article 73 — serious-incident reporting
EU AI Act · Article 73
Serious-incident reporting clock
Observation
What counts · Article 3(49)
A serious incident is an incident or malfunctioning that directly or indirectly leads to any of:
(a)Death or serious harm to health
(b)Serious + irreversible disruption of critical infrastructure
(c)Infringement of fundamental-rights obligations under Union law
(d)Serious harm to property or environment
Notification
Deployer chain · Article 26(5)
The deployer is usually the first to observe. Inform in order:
Deployer
→
Provider first
→
Importer / Distributor
+
Market-surveillance authority
FallbackIf the provider cannot be reached, Article 73 applies mutatis mutandis — the deployer takes over the reporting duty.
Reporting clock
Three deadlines · Article 73(2)–(4)
Each deadline has its own trigger and its own outer cap. "Immediately" runs alongside the cap in every case — the cap is a backstop, not the only constraint.
Article 73(2)
Default serious incident: 3(49)(a) (other than death), (c), (d). Trigger: established causal link or reasonable likelihood.
Immediately≤ 15 days after becoming aware
Article 73(3)
Widespread infringement or serious + irreversible disruption of critical infrastructure · 3(49)(b). Trigger: provider/deployer becomes aware.
Immediately≤ 2 days after becoming aware
Article 73(4)
Death of a person. Trigger: established or suspected causal relationship.
Immediately≤ 10 days after becoming aware
Authority handoffs
Downstream chain · Article 73(7), (8), (11)
Article 73(7)
For 3(49)(c) fundamental-rights infringement: the MSA informs the Article 77(1) public authorities.
Article 73(8)
MSA takes appropriate measures under Regulation (EU) 2019/1020 Article 19, within 7 days of notification.
Article 73(11)
National competent authority immediately notifies the Commission under Reg. 2019/1020 Article 20.
Article 73(1) — what to report
Providers of high-risk AI systems placed on the Union market shall report any serious incident to the market surveillance authorities of the Member States where that incident occurred.
— Article 73(1), Regulation (EU) 2024/1689
The provider is the Article 73(1) reporting addressee. The deployer-side serious-incident escalation duty is in Article 26(5), not in Article 73(1): the deployer must immediately inform first the provider, and then the importer or distributor and the relevant market-surveillance authority of any serious incident. If the deployer cannot reach the provider, Article 73 applies mutatis mutandis — the deployer takes over the reporting duty. The two duties run together — deployer observation triggers the provider's Article 73 reporting clock.
Article 73(2)–(4) — deadlines, by event severity
The report referred to in paragraph 1 shall be made immediately after the provider has established a causal link between the AI system and the serious incident or the reasonable likelihood of such a link, and, in any event, not later than 15 days after the provider or, where applicable, the deployer, becomes aware of the serious incident.
The period for the reporting referred to in the first subparagraph shall take account of the severity of the serious incident.
— Article 73(2), Regulation (EU) 2024/1689
Notwithstanding paragraph 2 of this Article, in the event of a widespread infringement or a serious incident as defined in Article 3, point (49)(b), the report referred to in paragraph 1 of this Article shall be provided immediately, and not later than two days after the provider or, where applicable, the deployer becomes aware of that incident.
— Article 73(3), Regulation (EU) 2024/1689
Notwithstanding paragraph 2, in the event of the death of a person, the report shall be provided immediately after the provider or the deployer has established, or as soon as it suspects, a causal relationship between the high-risk AI system and the serious incident, but not later than 10 days after the date on which the provider or, where applicable, the deployer becomes aware of the serious incident.
— Article 73(4), Regulation (EU) 2024/1689
| Event | Deadline |
|---|---|
| Default — serious incident under Article 3(49)(a) (other than death), (c), (d) | Immediately + not later than 15 days after becoming aware (Article 73(2)) |
| Widespread infringement OR serious + irreversible disruption of critical infrastructure (Article 3(49)(b)) | Immediately + not later than 2 days after becoming aware (Article 73(3)) |
| Death of a person | Immediately + not later than 10 days after becoming aware (Article 73(4)) |
The "immediately" duty runs alongside the outer cap in every case — the cap is a backstop, not the only constraint. Reporting starts as soon as the causal link is established (or, for death, suspected).
Article 73(5) allows the provider to submit an initial, possibly incomplete report when the full assessment is not yet available, with completion of the report as soon as the assessment is complete. Article 73(6) imposes investigation duties on the provider, including a risk assessment of the incident and corrective action, and requires cooperation with competent authorities; the provider does not destroy evidence relevant to the investigation.
Article 73(7)–(11) — fundamental-rights authority notification, MSA action, sectoral integration, Commission notification
- Article 73(7) — where the serious incident involves the infringement of obligations under Union law intended to protect fundamental rights (Article 3(49)(c)), the market-surveillance authority shall inform the national public authorities or bodies referred to in Article 77(1) (the public authorities or bodies that supervise or enforce the relevant fundamental-rights obligations). The Commission is to develop dedicated guidance to facilitate Article 73 compliance.
- Article 73(8) — upon receiving a notification of a serious incident, the market-surveillance authority shall take appropriate measures in accordance with Article 19 of Regulation (EU) 2019/1020, within seven days of receiving the notification.
- Article 73(9) — for high-risk AI systems referred to in Annex III placed on the market or put into service by providers subject to Union legislative instruments laying down equivalent reporting obligations, the notification of serious incidents shall be limited to Article 3(49)(c) (infringement of fundamental-rights obligations).
- Article 73(10) — for high-risk AI systems that are safety components of devices, or are themselves devices, covered by the MDR (Regulation (EU) 2017/745) or the IVDR (Regulation (EU) 2017/746), the notification of serious incidents shall be limited to Article 3(49)(c) and reported to the national competent authority chosen for that purpose by the Member States where that incident occurred (interface with MDR Article 87 / IVDR Article 82 vigilance regimes).
- Article 73(11) — the national competent authorities shall immediately notify the Commission of any serious incident, whether or not they have taken action on it, in accordance with Article 20 of Regulation (EU) 2019/1020.
The GPAI distinction is anchored in Article 55(1)(c), not in Article 73(11): for systemic-risk GPAI models, the provider reports serious incidents and possible corrective measures to the AI Office and, as appropriate, to national competent authorities. The Article 55(1)(c) report is at the model level and goes to the AI Office; the Article 73 report is at the system level and goes to the market-surveillance authority. Where a high-risk AI system is built on a systemic-risk GPAI model, both duties apply to the respective providers; the downstream system provider's Article 73 duty is not discharged by the upstream GPAI provider's Article 55(1)(c) report.
Article 73 ≠ Article 33 GDPR — the high-frequency confusion
Two distinct duties on a single incident
A single event can trigger both Article 73 EU AI Act and Article 33 GDPR. The duties run in parallel.
| Dimension | Article 73 EU AI Act | Article 33 GDPR |
|---|---|---|
| Trigger event | Serious incident under Article 3(49) — death, serious harm to health, serious + irreversible disruption of critical infrastructure, infringement of fundamental-rights obligations, serious harm to property or environment | Personal-data breach (Article 4(12) GDPR) |
| Addressee | Provider of a high-risk AI system (deployer informs provider) | Controller |
| Recipient | Market-surveillance authority of the Member State where the incident occurred | Competent supervisory authority |
| Deadline (baseline) | Immediately + not later than 15 days | 72 hours |
| Deadline (severe) | Immediately + not later than 10 days (death) / immediately + not later than 2 days (widespread or critical-infrastructure disruption) | 72 hours (uniform) |
| Risk standard | Causal link or reasonable likelihood of causal link with the AI system | Risk to rights and freedoms of natural persons |
| Recipient if low-risk | Always reported if Article 3(49) threshold met | No notification if unlikely to result in a risk to natural persons |
A high-risk AI system that suffers a data exfiltration causing personal-data breach and infringes fundamental rights triggers both. Report each independently.
How to operationalise Articles 72 / 73 in Modulos
The post-market and serious-incident regime maps onto these MFF-1 / OFF-1 requirements:
| Requirement | Description | OJ Article |
|---|---|---|
MRF-46 | Post-Market Monitoring System | Article 72 |
MRF-1, ORF-1 | Risk Management System (app / org) | Article 9 (feedback loop from PMM via Article 9(2)(c)) |
MRF-10 | Keeping Logs by Provider | Article 12 / 19 |
MRF-50 | Keeping Logs by Deployer | Article 26(6) |
ORF-11 | Corrective Actions (org) | Article 20 |
ORF-12 | Duty of Information (org) | Article 20 |
MRF-14, ORF-14 | Use and Oversight (deployer) | Article 26 |
ORF-49 | Deployment Monitoring and Incident Handling (org) | Article 26 |
MRF-49 | Deployment Monitoring and Incident Handling (app) | Article 26(5) |
ORF-47 | Serious Incident Reporting for Developers (org) | Article 73 (provider side) |
ORF-97 | GPAIM-SR Serious Incident Reporting (org) | Article 55(1)(c) (GPAI to AI Office — distinct from Article 73) |
Operating rules:
- Article 72 post-market monitoring plan (
MRF-46) — the plan content (Annex IV point 9) is the provider's authoring task; Modulos backs the plan with control-level evidence and links it to the rest of the technical documentation. - Continuous monitoring signals can be ingested via Modulos Runtime Inspection (Sources → Tests → Schedules → Results). Test failures surface as signals that feed the Article 72 analysis and, where the threshold is met, the Article 73 reporting evaluation.
- Article 73 serious-incident reports (
ORF-47) — the report content is the provider's authoring task; the report submission to the market-surveillance authority is the provider's action, not via Modulos. The submission confirmation and the report artefact live as versioned control-level evidence. - Article 26(5) / (6) deployer duties (
MRF-49,MRF-50,ORF-49) — Modulos records deployer-side log retention attestation (Article 26(6)), the deployer's incident-observation timestamp, and the upstream notification to the provider (Article 26(5)) as control-level evidence. - Article 20 corrective actions (recall, withdrawal, disablement) on
ORF-11/ORF-12— actions are recorded as control-level evidence on the affected MFF-1 requirements; the action's effect on conformity is reflected in the requirement's readiness signal. - Article 9 risk-management feedback loop — post-market data feeds Article 9(2)(c), recorded as an evidence update on
MRF-1/ORF-1controls. - Article 55(1)(c) systemic-risk GPAI reports (
ORF-97) report to the AI Office, distinct from the system-level Article 73 reports (ORF-47) that go to market-surveillance authorities.
Platform Article-numbering caveat: MRF-46 (post-market monitoring) references Article 61 internally; ORF-47 (serious-incident reporting) references Article 62. The final OJ-published Articles are 72 (PMM) and 73 (serious-incident) respectively.
Cross-framework mapping (preview)
| EU AI Act | Adjacent provision |
|---|---|
| Article 72 post-market monitoring | NIST AI RMF MANAGE 4; ISO 42001 Clause 9 (performance evaluation) and Clause 10 (improvement); ISO 13485 post-market surveillance (medical devices) |
| Article 73 serious-incident reporting | Article 33 GDPR personal-data breach (distinct duty); NIS2 Article 23 incident notification (different threshold, different recipient); MDR Article 87 medical-device vigilance |
| Article 3(49) serious-incident definition | NIS2 Article 23 'significant incident' definition (different scope); GDPR Article 4(12) 'personal data breach' (different scope) |
| Article 20 corrective actions | ISO 9001 Clause 10.2 nonconformity; product-safety recall procedures under sectoral law |
| Article 55(1)(c) systemic-risk GPAI reports | NIST AI RMF MEASURE / MANAGE; Article 73 EU AI Act (distinct — different recipient, different level) |
| Article 26(6) log retention | GDPR Article 5(1)(e) storage limitation (different basis); sectoral log-retention regimes |
Related pages
High-risk AI systems
Article 9 risk management; Article 12 logging — the loop that feeds Article 72
Conformity assessment and CE marking
Article 43(4) — substantial-modification triggers re-assessment
Roles and responsibilities
Article 26(5) / (6) deployer duties; Article 22 authorised-representative cooperation
General-purpose AI models
Article 55(1)(c) systemic-risk GPAI serious-incident reporting to the AI Office
Operationalizing in Modulos
Evidence patterns for PMM plan, incident artefacts, corrective actions
Source attribution
Regulation (EU) 2024/1689 — Articles 3(49), 20, 26(5)–(6), 55, 72, 73, 79 and Annex IV — is published in the Official Journal of the European Union L of 12 July 2024 (CELEX 32024R1689). Verbatim blockquotes on this page reflect the OJ-published text. Regulation (EU) 2016/679 (GDPR), Regulation (EU) 2017/745 (MDR), and Regulation (EU) 2017/746 (IVDR) are referenced. The AI Omnibus consolidated text, when published, will supersede where it amends.
Disclaimer
This page is for general informational purposes and does not constitute legal advice.