EU AI Act Art. 15 and robot AI policies

From the regulatory clock

At a glance

Read from the site’s regulatory clock, the record that is re-verified against primary law, as of 19 September 2026.

Instrument: Regulation (EU) 2024/1689 — general application, transparency, governance and penalties — Binding law (for the actor it names)
Applies?: Applies since 2 August 2026 — in application as of 19 September 2026
Responsible actor: Provider / deployer as applicable per obligation.
What Provael measures toward it: Nothing directly — this entry exists so the clock is not misread as 'the AI Act does not apply yet'.
What that does not establish: Not a compliance position on any Article 50 obligation.
Source and review: Primary source · verified 19 September 2026

Instrument: Regulation (EU) 2024/1689 - Article 15 (accuracy, robustness, cybersecurity) — Binding law (for the actor it names)
Applies?: Applies from 2 August 2028 — upcoming as of 19 September 2026
Responsible actor: Provider of the high-risk AI system / product manufacturer (and a notified body where a third-party route applies).
What Provael measures toward it: A measured attack-success rate with a 95% Wilson CI and a benign control is candidate evidence toward Art. 15 robustness documentation.
What that does not establish: Does not establish conformity, is not a certificate, and is not a notified-body opinion.
Source and review: Primary source · verified 19 September 2026

Use it for

What a Provael report can be used for

Robustness

High-risk AI must be resilient to errors, faults and inconsistencies, and to malicious third-party attempts to alter use or performance by exploiting vulnerabilities.

Declared metrics

Accuracy and the relevant accuracy metrics must be declared. An attack-success rate with a confidence interval and a benign control is a robustness figure that can be declared the same way.

The evidence

An attack-success rate with a 95% Wilson confidence interval and a benign false-positive control: a declared robustness metric, and candidate evidence toward Art. 15 documentation.
A reproducible attack trace per finding, so the robustness evidence is auditable.
A SARIF report and CI red-team gate that demonstrate ongoing robustness testing across the lifecycle.

It also maps to

Machinery Regulation Annex III 1.1.9 and 1.2.1: For AI-enabled machinery these are the requirements that apply, from 20 January 2027, in place of a direct Article 15 duty: protection against corruption, and control systems that withstand reasonably foreseeable malicious attempts and keep self-evolving behaviour within its safety envelope. The measured keep-out violation rate, its interval and the paired benign control are candidate evidence toward both, in the technical file. The full mapping is on /compliance/eu-machinery-regulation/.
Product Liability Directive (EU) 2024/2853: Software and AI systems are products; for products placed on the market after 9 December 2026 (transposition by the same date), defectiveness may be presumed where a manufacturer fails to disclose relevant evidence. A dated, reproducible robustness measurement is relevant evidence of the kind that duty covers. Not a defence and not a determination of defectiveness.
Article 9: Risk-management system - the measured ASR and its trace feed the identification and evaluation of reasonably foreseeable misuse.
Article 72: Post-market monitoring - the CI red-team gate is a monitoring signal that re-tests robustness on every checkpoint after release.
Article 11 / Annex IV: Technical documentation - carried by the CycloneDX ML-BOM Provael emits for the policy under test.

Limits

What it cannot be used for

A mapping identifies possible relevance. It does not establish that this framework applies to your product, that anything here is compliance, that an assessor or insurer will accept it, or a real-world risk probability. Whether it applies to your product depends on your role, market and product class: a question for your counsel, not a determination this page makes.

Not, on its own, an Article 15 position for machinery: since 27 July 2026 Article 15 reaches AI-enabled machinery only through Machinery Regulation Annex III, so that is the text a robot policy’s evidence is filed against.
Simulation evidence for the tested checkpoint and protocol. Physical safety and conformity require separate assessment.
Provael measures adversarial robustness - susceptibility to manipulation - not general accuracy, reliability, or functional safety.
The measured result uses templated, auditable attacks; the search-based families (optimized, universal_patch, gradient_patch) have no real-policy rate. Results are a floor on susceptibility - a behavioural lower bound, not a certified worst-case bound.

Evidence, not certification

Running Provael does not make a system compliant or certified - it generates measurements you can put into a conformity or assurance file. Provael is not a notified body, a lab, or a certification scheme.

Independent project. Not affiliated with or endorsed by ISO, the EU, NIST, IEC, OWASP, or MITRE. Not legal advice.

Clause references are indicative: check each against the text you hold before you cite it.

What it is

The framework

Article 15 sits in the high-risk requirements of the EU AI Act (Regulation (EU) 2024/1689): high-risk AI systems must perform consistently across their lifecycle on accuracy, robustness and cybersecurity, with the relevant metrics declared in the accompanying documentation, and must be resilient against adversarial manipulation - data poisoning, model poisoning, adversarial examples, and attempts to exploit the system to alter its behaviour.
What changed on 27 July 2026: Regulation (EU) 2026/1744, the Digital Omnibus on AI, entered into force and moved the Machinery Regulation from AI Act Annex I Section A to Section B (point 21). For AI-enabled machinery, Chapter III - Article 15 included - therefore does not apply directly. For a robot policy that is a safety component or is embedded in machinery, the text that applies is Machinery Regulation Annex III - essential requirements 1.1.9 (protection against corruption) and 1.2.1 (safety and reliability of control systems, including self-evolving behaviour and reasonably foreseeable malicious attempts) - amended by delegated acts under Article 8(3) that must apply by 2 August 2028 and are to reflect Article 15.
Between 20 January 2027 and those delegated acts, Machinery Regulation Article 20(10) is the bridge: an AI system that complies with the harmonised standards or common specifications adopted under AI Act Articles 40 and 41 is presumed to conform with the Annex III requirements for high-risk AI. No such harmonised standard had been cited in the Official Journal when this page was last verified, so the presumption has nothing to borrow yet.
Direct application of Article 15 is for the products that stayed in Annex I Section A and for the stand-alone high-risk systems of Annex III, from 2 December 2027 (Annex III) and 2 August 2028 (Annex I Section A). The nearer binding date for a robot policy is the Machinery Regulation itself, 20 January 2027, with the revised Product Liability Directive applying to products placed on the market after 9 December 2026.
No harmonised robustness standard exists yet under either regulation (CEN-CENELEC JTC 21 targets Q4 2026 for the AI Act; the functional-safety texts for AI elements, ISO/IEC TS 22440-1/-2/-3, were at Committee Draft with publication expected in 2027). Provael’s Art. 15 example uses ISO/IEC TR 24029 as its empirical-robustness methodology anchor and also cites TS 22440.

Timing

Dates (verified 19 September 2026)

The Digital Omnibus on AI in force - Machinery Regulation moved to Annex I Section B: 27 July 2026Regulation (EU) 2026/1744: Parliament 16 June 2026, Council 29 June 2026, published OJ 24 July 2026, in force on the third day after publication. Point (41)(b) adds the Machinery Regulation to Annex I Section B as point 21; from this date Chapter III does not apply directly to AI-enabled machinery.
Revised Product Liability Directive applies to products placed on the market: 9 December 2026Directive (EU) 2024/2853, Article 2(1); transposition by the same date (Article 22(1)). Software and AI systems are products.
Machinery Regulation applies: 20 January 2027Regulation (EU) 2023/1230, Article 54 as corrected by the corrigendum of 4 July 2023. The nearer binding date for a robot policy; Article 20(10) bridges the AI requirements until the delegated acts below.
Article 15 applies directly to stand-alone Annex III high-risk systems: 2 December 2027Regulation (EU) 2026/1744, point (40), replacing Article 113, third paragraph, point (a).
Article 15 applies directly to Annex I Section A products; Machinery Annex III delegated acts apply by: 2 August 2028Point (40) of the same Regulation for Annex I Section A; for machinery the requirements arrive by Commission delegated acts under Machinery Regulation Article 8(3), applicable by this date. Re-read on EUR-Lex 19 September 2026. Verify the primary source before relying on it.
Superseded statutory date (product-embedded high-risk, Annex I): 2 August 2027The original date under Regulation (EU) 2024/1689, kept here so a reader who learned it knows what replaced it.

Check the live EUR-Lex or ISO text before relying on a date.

Sources

Primary references

Sample SARIF ↓All frameworks

See what a report contains.

The sample report comes from a public run and is published in full, nothing redacted. To have this mapping filled in for your own policy, request a scoping call.

See the sample report Request a scoping call