Art. 26: Obligations of deployers of high-risk AI systems under the EU AI Act
Art. 26 requires the deployer of a high-risk AI system to use it in accordance with the instructions for use, to assign human oversight to persons with competence, training, authority and support, to keep input data under its control representative, to monitor operation and report risks and serious incidents, to retain logs, and to inform workers and affected natural persons. In this mapping, eight elements are testable, eight are documentation and eleven are process. Regulation (EU) 2026/1744 moved the Annex III application dateDeferred and left the content of Art. 26 unchanged.
Art. 26 builds on three provider outputs: the instructions for use under Art. 13, the human oversight measures the provider identifies under Art. 14(3), and the logs the system generates under Art. 12. Art. 26(5) refers to Art. 72 for post-market monitoring, Art. 73 for serious incident reporting and Art. 79(1) for the meaning of a risk that triggers suspension of use. A deployer covered by Art. 27 also carries out a fundamental rights impact assessment.
What the text requires and how each element is verified
Requires
Testable
Documentation
Process
Art. 26(1)Deployers of high-risk AI systems take appropriate technical and organisational measures. The measures make sure that the deployer uses the system in accordance with the instructions for use that accompany the system. This duty applies pursuant to paragraphs 3 and 6.
Art. 26(2)Deployers assign human oversight to natural persons. These persons have the necessary competence, training and authority. These persons also have the necessary support.
Art. 26(3)The obligations in paragraphs 1 and 2 are without prejudice to other deployer obligations under Union or national law. They are also without prejudice to the deployer's freedom to organise its own resources and activities to implement the human oversight measures indicated by the provider.
Art. 26(4)This paragraph is without prejudice to paragraphs 1 and 2. To the extent the deployer exercises control over the input data, the deployer makes sure that the input data is relevant and sufficiently representative in view of the intended purpose of the high-risk AI system.
Art. 26(5)Deployers monitor the operation of the high-risk AI system on the basis of the instructions for use. Where relevant, they inform providers in accordance with Art. 72. Where deployers have reason to consider that use in accordance with the instructions for use may result in the system presenting a risk within the meaning of Art. 79(1), they inform the provider or distributor and the relevant market surveillance authority without undue delay, and they suspend the use of the system. Where deployers have identified a serious incident, they immediately inform first the provider, then the importer or distributor, and the relevant market surveillance authorities. If the deployer cannot reach the provider, Art. 73 applies mutatis mutandis. This obligation does not cover sensitive operational data of deployers that are law enforcement authorities. A deployer that is a financial institution subject to internal governance requirements under Union financial services law meets the monitoring obligation through the rules on internal governance arrangements, processes and mechanisms under that law.
Art. 26(6)Deployers keep the logs automatically generated by the high-risk AI system to the extent such logs are under their control. They keep the logs for a period appropriate to the intended purpose of the system, of at least six months. Applicable Union or national law, in particular Union law on the protection of personal data, may set a different period. Deployers that are financial institutions subject to internal governance requirements under Union financial services law maintain the logs as part of the documentation kept under the relevant Union financial services law.
Art. 26(7)Before putting into service or using a high-risk AI system at the workplace, deployers who are employers inform workers' representatives and the affected workers that they will be subject to the use of the high-risk AI system. Where applicable, this information follows the rules and procedures laid down in Union and national law and practice on information of workers and their representatives.
Art. 26(8)Deployers of high-risk AI systems that are public authorities, or Union institutions, bodies, offices or agencies, comply with the registration obligations in Art. 49. When such a deployer finds that the high-risk AI system it envisages using is not registered in the EU database referred to in Art. 71, it does not use that system and informs the provider or the distributor.
Art. 26(9)Where applicable, deployers of high-risk AI systems use the information provided under Art. 13 to comply with their obligation to carry out a data protection impact assessment under Art. 35 of Regulation (EU) 2016/679 or Art. 27 of Directive (EU) 2016/680.
Art. 26(10)This paragraph is without prejudice to Directive (EU) 2016/680. This paragraph covers an investigation for the targeted search of a person suspected or convicted of a criminal offence. The deployer of a high-risk AI system for post-remote biometric identification requests an authorisation for the use of that system. The authorisation comes from a judicial authority or an administrative authority whose decision is binding and subject to judicial review. The request is made ex ante, or without undue delay and no later than 48 hours. The exception is use for the initial identification of a potential suspect based on objective and verifiable facts directly linked to the offence. Each use is limited to what is strictly necessary for the investigation of a specific criminal offence. If the authorisation is rejected, use of the system linked to that request stops with immediate effect, and the personal data linked to that use is deleted. Such a system is never used for law enforcement purposes in an untargeted way. Use requires a link to a criminal offence, a criminal proceeding, a genuine and present or genuine and foreseeable threat of a criminal offence, or the search for a specific missing person. No decision that produces an adverse legal effect on a person is taken by law enforcement authorities based solely on the output of such a system. This paragraph is without prejudice to Art. 9 of Regulation (EU) 2016/679 and Art. 10 of Directive (EU) 2016/680 for the processing of biometric data. Each use is documented in the relevant police file and made available to the market surveillance authority and the national data protection authority on request, excluding sensitive operational data. Deployers submit annual reports on their use of such systems to those authorities, excluding sensitive operational data; reports may be aggregated across deployments. Member States may introduce more restrictive laws on the use of such systems in accordance with Union law.
Art. 26(11)This paragraph is without prejudice to Art. 50. Deployers of high-risk AI systems referred to in Annex III that make decisions or assist in making decisions related to natural persons inform the natural persons that they are subject to the use of the high-risk AI system. For high-risk AI systems used for law enforcement purposes, the information duty is the one in Art. 13 of Directive (EU) 2016/680.
Art. 26(12)Deployers of high-risk AI systems cooperate with the relevant competent authorities in any action those authorities take in relation to the high-risk AI system in order to implement the Regulation.
The deployed configuration, the input types and the use cases fall within the intended purpose and the conditions of use stated in the provider's instructions for use (Art. 26(1))
The human oversight measures indicated by the provider under Art. 14(3)(b) are implemented in the deployment and can be exercised: stop, override, disregard the output, and intervene on the operation (Art. 26(2) and 26(3))
The natural persons assigned to human oversight hold the authority to act: a stop or override instruction from an assigned person takes effect on the system in a test (Art. 26(2))
The assigned persons have the necessary support in the deployed workflow: access to the instructions for use, the output information the provider declares under Art. 13, and an escalation path, verified in a walkthrough (Art. 26(2))
Where the deployer controls the input data, the live input population and feature distributions are relevant and sufficiently representative for the intended purpose declared in the instructions for use (Art. 26(4))
The suspension of use in Art. 26(5) can be executed: a suspension instruction halts the use of the system in the deployment in a test (Art. 26(5))
The logs automatically generated by the system and under the deployer's control are complete and readable: a sampled decision can be reconstructed from the retained logs (Art. 26(6))
The retention configuration of the log store enforces the period of at least six months in Art. 26(6), or the different period that applicable Union or national law sets (Art. 26(6))
The instructions for use received from the provider, with the intended purpose and the conditions of use identified (Art. 26(1))
The assignment record naming the natural persons who hold human oversight, with their competence, training and authority recorded (Art. 26(2))
The log retention policy stating the retention period under Art. 26(6), the scope of logs under the deployer's control, and, for financial institutions, the place of the logs in the documentation kept under Union financial services law (Art. 26(6))
The record that workers' representatives and the affected workers were informed before the system was put into service or used at the workplace (Art. 26(7))
The EU database registration record under Art. 49 for a deployer that is a public authority or a Union institution, body, office or agency (Art. 26(8))
The data protection impact assessment under Art. 35 of Regulation (EU) 2016/679 or Art. 27 of Directive (EU) 2016/680 that uses the Art. 13 information (Art. 26(9))
For post-remote biometric identification: the authorisation requests, the police file record of each use, and the annual reports to the market surveillance and data protection authorities (Art. 26(10))
The notice text that informs natural persons that they are subject to the use of the high-risk AI system (Art. 26(11))
Technical and organisational measures for use in accordance with the instructions for use are maintained over the life of the deployment, including after a provider update to the instructions (Art. 26(1))
Training and support for the persons assigned to human oversight are maintained and renewed when the system or the assigned persons change (Art. 26(2))
Where the deployer controls the input data, the input data is checked against the intended purpose at each change of data source or population (Art. 26(4))
The operation of the system is monitored on the basis of the instructions for use, and the provider is informed under Art. 72 where relevant (Art. 26(5))
A risk within the meaning of Art. 79(1) triggers notification of the provider or distributor and the market surveillance authority without undue delay and suspension of use (Art. 26(5))
A serious incident triggers immediate notification of first the provider, then the importer or distributor and the market surveillance authorities, with Art. 73 applied where the provider cannot be reached (Art. 26(5))
For financial institutions, the monitoring duty is fulfilled through the internal governance arrangements, processes and mechanisms under Union financial services law (Art. 26(5))
A deployer that is a public authority checks registration in the EU database before use, does not use an unregistered system, and informs the provider or the distributor (Art. 26(8))
For post-remote biometric identification: authorisation is requested ex ante or no later than 48 hours, use stops and personal data is deleted on rejection, each use is limited to what is strictly necessary, and no adverse legal decision is based solely on the output (Art. 26(10))
The notice to natural persons is delivered at the point where a decision is made or assisted (Art. 26(11))
The deployer cooperates with competent authorities in any action they take in relation to the system (Art. 26(12))
Metrics and where thresholds come from
Threshold sources, in order of precedence
A harmonised standard cited in the Official Journal (none as of the anchor date)
A JTC 21 draft standard (prEN), with its stage
An ISO/IEC standard or technical report, with its stage
A Commission guideline or code of practice, with its date from the standards data
NIST AI RMF or a sector framework
The state of the art, which the provider justifies and documents
The Act sets no numeric performance threshold for Art. 26. It sets two periods: under Art. 26(6) the deployer keeps the logs under its control for at least six months unless applicable Union or national law provides otherwise, and under Art. 26(10) a deployer of a post-remote biometric identification system requests authorisation ex ante or no later than 48 hours. For oversight competence, monitoring frequency and input data representativeness, no numeric threshold exists. No harmonised standard is cited in the Official Journal. The level in use today is the JTC 21 draft standards, prEN 18229-3 (draft) and prEN 18229-1 (under enquiry), together with the provider's instructions for use, which are the deployer's reference under Art. 26(1). The deployer justifies and documents its targets against the context of use; the acceptance criterion step in PR-14-OVR and PR-12-LOG records that justification.
Metric sources and their stage
Source and stage
Title
What it provides here
Verified on
prEN 18229-3Draft
AI trustworthiness framework, part 3: transparency and human oversight
draft content of the instructions for use under Art. 13 and of the human oversight measures under Art. 14 that the deployer follows under Art. 26(1) to (3); at drafting stage, no text is public
requirements under enquiry for what an Art. 12 log records and how it is retained, giving the Art. 26(6) retention and reconstruction check its expected event set and fields
Each protocol is versioned; its page states the derivation from the standards above. PR-14-OVR covers use within the instructions for use and the oversight, authority, support and suspension checks; PR-12-LOG covers log retention and decision reconstruction; PR-10-BIAS profiles the input data the deployer controls.
Instructions-for-use conformance record (PR-14-OVR output)
Human oversight assignment record (Art. 26(2))
Human oversight verification report (PR-14-OVR output)
Log retention policy and retention configuration evidence (Art. 26(6))
Decision reconstruction record (PR-12-LOG output)
Input data representativeness profile (PR-10-BIAS output)
Worker information record (Art. 26(7))
Notice to natural persons subject to the use of the high-risk AI system (Art. 26(11))
Common findings
Human oversight is assigned on an organisation chart without a record of the competence, training or authority of the named persons.
The stop or override function exists in the provider's product but is not reachable by the assigned overseer in the deployed configuration.
Logs are generated on the provider's platform and the retention period for the logs under the deployer's control is unknown or shorter than the period in Art. 26(6).
The system is used for a purpose or a population outside the conditions of use stated in the instructions for use.
Workers and affected natural persons are informed after the system is in use rather than before, or not at all.
The monitoring duty has no defined route from an observed risk to the suspension of use and the notification of the provider and the market surveillance authority.
Questions
+−What are the EU AI Act Article 26 deployer obligations?
Art. 26 lists twelve paragraphs of duties for deployers of high-risk AI systems. The operational duties are use in accordance with the instructions for use under Art. 26(1), human oversight assigned to competent persons under Art. 26(2), input data relevance and representativeness under Art. 26(4), monitoring and incident reporting under Art. 26(5), and log retention under Art. 26(6). The information duties are to workers under Art. 26(7) and to natural persons subject to the system under Art. 26(11). Public authorities register the system under Art. 26(8) and Art. 49. Regulation (EU) 2026/1744 did not change the content of Art. 26.
+−What is the EU AI Act deployer definition?
Art. 3(4) of Regulation (EU) 2024/1689 defines a deployer as a natural or legal person, public authority, agency or other body that uses an AI system under its authority. Use in the course of a personal non-professional activity is excluded. Art. 26 applies to a deployer of a high-risk AI system. A deployer becomes the provider of a high-risk AI system in the cases listed in Art. 25(1).
+−AI Act deployer or provider: when does a deployer take on provider duties?
Art. 25(1) names three cases. The deployer puts its name or trademark on a high-risk AI system already placed on the market or put into service. The deployer makes a substantial modification to a high-risk AI system that remains high-risk. The deployer modifies the intended purpose of an AI system so that it becomes high-risk. In each case the deployer is considered a provider and carries the provider obligations under Art. 16. The initial provider then supplies the technical documentation and access set out in Art. 25(2).
+−When do the EU AI Act deployer obligations apply?
Art. 26 sits in Chapter III Section 3 of Regulation (EU) 2024/1689. Amended Art. 113(c) applies that section from the Annex III application date for systems classified under Art. 6(2) and from the Annex I application date for systems classified under Art. 6(1). Regulation (EU) 2026/1744 moved these dates and left the content of Art. 26 unchanged.
+−Which deployer obligations under the AI Act can be tested?
Four groups of elements in Art. 26 produce an observable result. Use within the instructions for use under Art. 26(1) is checked against the provider's declared intended purpose and conditions of use. The oversight measures under Art. 26(2) and 26(3) are exercised in a test: stop, override and disregard the output. Input data under the deployer's control is profiled against the intended purpose under Art. 26(4). Log retention and decision reconstruction are verified under Art. 26(6). The other paragraphs are met by a document or by a repeated organisational activity.
+−What connects to Article 26 of the AI Act?
Art. 26(1) relies on the instructions for use that the provider supplies under Art. 13. Art. 26(2) and 26(3) implement the human oversight measures the provider identifies under Art. 14(3)(b). Art. 26(6) retains the logs the system generates under Art. 12. Art. 26(5) points to Art. 72 for post-market monitoring, Art. 73 for serious incident reporting and Art. 79(1) for the meaning of risk. Art. 26(8) points to Art. 49 for registration and Art. 71 for the EU database. Art. 26(9) points to the data protection impact assessment under Art. 35 of Regulation (EU) 2016/679. Art. 27 adds the fundamental rights impact assessment for certain deployers. Art. 26(11) is without prejudice to Art. 50.
How we test this
We test the deployed configuration against the intended purpose and conditions of use in the provider's instructions under Art. 26(1), we exercise the stop, override and disregard functions from the assigned overseer's position under Art. 26(2), and we verify that the logs under the deployer's control reconstruct a sampled decision and that the retention configuration enforces the Art. 26(6) period. We document the results as the artifacts listed above; the scope limits are stated on what we cannot do.
We do not perform conformity assessments where a notified body is required.
We do not issue CE marking or the EU Declaration of Conformity.
We are not a notified body.
We do not give legal opinions on classification or liability.
We test against requirements, prepare evidence, and structure documentation; compliance decisions remain with the provider or deployer, their notified body, and their counsel.