Skip to main content

Article 15 for HR Tech: Accuracy, Robustness and Cybersecurity

How Article 15 (accuracy, robustness, cybersecurity) applies to recruitment and worker-management AI under Annex III point 4: scope, roles, dates, fines and what the Act leaves open.

Providers of high-risk AI systems for recruitment or worker management must design them for appropriate accuracy, robustness and cybersecurity (Article 15).

These are Annex III point 4 systems, and Article 15 applies to them from 2 December 2027. The provider declares accuracy levels and metrics in the instructions for use (Article 15(3)). Employers that only use such a system are deployers and are not bound by Article 15. Fines reach EUR 15 million or, for an undertaking, 3% of worldwide annual turnover, whichever is higher (Article 99(4)). Next step: list the metrics you will declare.

Article 15 asks for appropriate performance, not a fixed number

Article 15 sets design requirements for the provider. It names no accuracy value for any sector.

  • Art. 15(1)
    Appropriate level, throughout the lifecycle

    High-risk AI systems must be designed and developed to reach an appropriate level of accuracy, robustness and cybersecurity, and to perform consistently in those respects throughout their lifecycle.

  • Art. 15(2)
    Benchmarks are for the Commission to encourage

    The Commission is to encourage, as appropriate, benchmarks and measurement methodologies, in cooperation with stakeholders such as metrology and benchmarking authorities. This is not a duty of the provider.

  • Art. 15(3)
    Declare levels and metrics

    The levels of accuracy and the relevant accuracy metrics must be declared in the accompanying instructions for use.

  • Art. 15(4)
    Resilience against errors, and feedback loops

    Systems must be as resilient as possible against errors, faults or inconsistencies in the system or its environment, in particular from interaction with natural persons or other systems. Technical and organisational measures must be taken; redundancy, backup or fail-safe plans are named as options. A system that keeps learning after it is placed on the market must be developed to eliminate or reduce as far as possible the risk that biased outputs influence future inputs (feedback loops).

  • Art. 15(5)
    Cybersecurity against unauthorised third parties

    Systems must be resilient against attempts by unauthorised third parties to alter their use, outputs or performance by exploiting vulnerabilities. Where appropriate, the technical solutions must include measures against data poisoning, model poisoning, adversarial examples or model evasion, confidentiality attacks and model flaws.

Which HR-tech AI must meet Article 15

Article 15 binds providers of high-risk AI systems only. The rows below show where HR-tech AI becomes high-risk.

  • HR tech is not a category in the Act

    The Act has no HR-tech category. This site uses the word for AI that reaches the Act through Annex III point 4 (employment, workers’ management and access to self-employment), and through the ban in Article 5(1)(f) on inferring emotions in the workplace.

  • Annex III point 4: employment and workers’ management

    Annex III point 4(a) lists AI systems intended to be used for the recruitment or selection of natural persons, in particular to place targeted job advertisements, to analyse and filter job applications, and to evaluate candidates. Point 4(b) lists AI systems intended to be used to make decisions affecting terms of work-related relationships, the promotion or termination of work-related contractual relationships, to allocate tasks based on individual behaviour or personal traits or characteristics, or to monitor and evaluate the performance and behaviour of persons in such relationships. An Annex III system is treated as not high-risk only under Article 6(3), and the provider must document that assessment before the system is placed on the market or put into service (Article 6(4)). Article 6(3) never applies where the system performs profiling of natural persons. A provider that concludes its Annex III system is not high-risk is subject to the registration obligation in Article 49(2).

  • Provider and deployer

    The provider, for example an HR-software vendor, develops the AI system, or has it developed, and places it on the market or puts it into service under its own name (Article 3(3)). An employer, recruitment agency or other body that uses a system under its authority is a deployer (Article 3(4)). A deployer becomes a provider if it puts its name or trademark on a high-risk system already on the market, makes a substantial modification that leaves the system high-risk, or changes the intended purpose of a non-high-risk system so that it becomes high-risk (Article 25(1)). The Act does not apply to the obligations of deployers who are natural persons using AI systems in a purely personal non-professional activity (Article 2(10)).

  • Dates

    Annex III point 4 systems must meet these duties from 2 December 2027 (Article 113, third paragraph, point (c)(i)). A system placed on the market or put into service before that date is covered only if its design changes significantly afterwards; in any case, providers and deployers of high-risk systems intended to be used by public authorities must take the necessary steps to comply by 2 August 2030 (Article 111(2)). The ban in Article 5(1)(f) has applied since 2 February 2025; the carve-out in Article 113, third paragraph, point (a) does not concern it.

  • Fines

    Breaches of provider obligations under Article 16 and of deployer obligations under Article 26 can be fined up to EUR 15 million or, if the offender is an undertaking, up to 3% of its total worldwide annual turnover in the preceding financial year, whichever is higher (Article 99(4)). For SMEs, including start-ups, each fine is up to whichever is lower (Article 99(6)); the same applies to small mid-cap enterprises (Article 3(14b), Article 99(6a)). Each Member State decides to what extent fines may be imposed on public authorities and bodies established there (Article 99(8)).

  • Scope limits

    The Act does not apply to areas outside the scope of Union law (first sentence of Article 2(3)), to AI systems or AI models, including their output, specifically developed and put into service for the sole purpose of scientific research and development (Article 2(6)), or to research, testing or development before a system is placed on the market or put into service; testing in real world conditions is not covered by that exclusion (Article 2(8)). It does not apply to AI systems released under free and open-source licences, unless they are placed on the market or put into service as high-risk AI systems or fall under Article 5 or 50 (Article 2(12)).

What the Act leaves open for HR tech

  • No accuracy value or fairness metric

    Article 15 names no accuracy metric, minimum value, test method or standard for any sector, and the Act names no fairness metric for HR systems. The provider declares the levels and relevant metrics (Article 15(3)). The technical documentation must also describe the metrics used to measure accuracy, robustness and compliance, as well as potentially discriminatory impacts (Annex IV point 2(g)).

  • Presumptions of conformity

    The Act names no standard itself. Systems in conformity with harmonised standards whose references are published in the Official Journal are presumed to comply with the Section 2 requirements to the extent the standards cover them (Article 40(1)). A cybersecurity certificate or statement of conformity under Regulation (EU) 2019/881, published in the Official Journal, gives a presumption for the cybersecurity requirements of Article 15 (Article 42(2)); systems within Regulation (EU) 2024/2847 that meet its Article 12(1) are deemed to comply with them (Article 42(3)). The Commission is to develop guidelines on the application of the requirements and obligations in Articles 8 to 15 and in Articles 25 and 26 (Article 96(1), point (a)).

  • Example: feedback loops in a ranking tool

    Example, not an Act requirement: a provider of a candidate-ranking tool that keeps learning after it is placed on the market could test whether its earlier rankings influence the inputs it later learns from, which is what Article 15(4) on feedback loops asks the provider to address.

Questions about Article 15 in HR tech

Does Article 15 set an accuracy value or a fairness metric for recruitment AI?

No. Article 15(1) requires an appropriate level, and Article 15(3) requires the provider to declare the levels and relevant accuracy metrics in the instructions for use. The Act names no value, metric or test method for HR systems.

Who must meet Article 15 when an employer buys a screening tool?

The provider designs the system to meet Article 15 (Article 16, point (a)). The employer is a deployer and uses the system according to the instructions for use (Article 26(1)). A deployer that puts its own name on the system, makes a substantial modification or changes the intended purpose becomes the provider (Article 25(1)).

From when does Article 15 apply to a CV-screening system?

From 2 December 2027, if the system is an Annex III point 4 system that is high-risk (Articles 6(2) and 113). A system already on the market before that date is covered only after a significant design change (Article 111(2)).

Related pages

Scanara prepares evidence from your repositories

Scanara reads your connected GitHub repositories and runs rule-based scans for standard AI and machine-learning patterns. It can miss non-standard code. Separately, it generates draft documentation from templates and your assessment answers as evidence for your compliance work. You make the classification decisions.