CHAPTER III HIGH-RISK AI SYSTEMS · SECTION 2 Requirements for high-risk AI systems
Article 15 Accuracy, robustness and cybersecurity
Navigation, not legal advice. These pages quote the consolidated text of the Act and link to it. They do not say whether the Act applies to you or what it requires of you; read the provisions in full, and ask a lawyer.
EUR-Lex: “This text is meant purely as a documentation tool and has no legal effect. The Union's institutions do not assume any liability for its contents. The authentic versions of the relevant acts, including their preambles, are those published in the Official Journal of the European Union and available in EUR-Lex. Those official texts are directly accessible through the links embedded in this document”
Article 15(1)
“High-risk AI systems shall be designed and developed in such a way that they achieve an appropriate level of accuracy, robustness, and cybersecurity, and that they perform consistently in those respects throughout their lifecycle.”
Article 15(2)
“To address the technical aspects of how to measure the appropriate levels of accuracy and robustness set out in paragraph 1 and any other relevant performance metrics, the Commission shall, in cooperation with relevant stakeholders and organisations such as metrology and benchmarking authorities, encourage, as appropriate, the development of benchmarks and measurement methodologies.”
Article 15(3)
“The levels of accuracy and the relevant accuracy metrics of high-risk AI systems shall be declared in the accompanying instructions of use.”
Article 15(4)
“High-risk AI systems shall be as resilient as possible regarding errors, faults or inconsistencies that may occur within the system or the environment in which the system operates, in particular due to their interaction with natural persons or other systems. Technical and organisational measures shall be taken in this regard.”
Article 15(4), second subparagraph
“The robustness of high-risk AI systems may be achieved through technical redundancy solutions, which may include backup or fail-safe plans.”
Article 15(4), third subparagraph
“High-risk AI systems that continue to learn after being placed on the market or put into service shall be developed in such a way as to eliminate or reduce as far as possible the risk of possibly biased outputs influencing input for future operations (feedback loops), and as to ensure that any such feedback loops are duly addressed with appropriate mitigation measures.”
Article 15(5)
“High-risk AI systems shall be resilient against attempts by unauthorised third parties to alter their use, outputs or performance by exploiting system vulnerabilities.”
Article 15(5), second subparagraph
“The technical solutions aiming to ensure the cybersecurity of high-risk AI systems shall be appropriate to the relevant circumstances and the risks.”
Article 15(5), third subparagraph
“The technical solutions to address AI specific vulnerabilities shall include, where appropriate, measures to prevent, detect, respond to, resolve and control for attacks trying to manipulate the training data set (data poisoning), or pre-trained components used in training (model poisoning), inputs designed to cause the AI model to make a mistake (adversarial examples or model evasion), confidentiality attacks or model flaws.”
Long paragraphs are cut at a word and end with an ellipsis; the link above has the full text.
“The technical solutions to address AI specific vulnerabilities shall include, where appropriate, measures to prevent, detect, respond to, resolve and control for attacks trying to manipulate the training data set (data poisoning), or pre-trained components used in training (model poisoning), inputs designed to cause the AI model to make a mistake (adversarial examples or model evasion), confidentiality attacks or model flaws.”
| In the Act | How the attack works | MITRE ATLAS |
|---|---|---|
| Data poisoning | No technique page matches | AML.T0020, AML.T0059 |
| Model poisoning | Supply chain and software flaws in AI tools | AML.T0018, AML.T0010 |
| Adversarial examples or model evasion | Jailbreaks and safety bypasses | AML.T0015, AML.T0043 |
| Confidentiality attacks | Model extraction and training data attacks, System prompt and hidden context leakage | AML.T0024 |
| Model flaws | No technique page matches | — |
Which pages match each class is ShieldMap's own reading, not the Act's. ATLAS techniques are also on atlas.mitre.org.
Extracts from the consolidated text 02024R1689-20260727 of Regulation (EU) 2024/1689 (02024R1689 — EN — 27.07.2026 — 001.001), read from EUR-Lex on 2026-11-19, which includes Regulation (EU) 2026/1744. Original act: OJ L, 2024-07-12. © European Union, https://eur-lex.europa.eu; reused under Commission Decision 2011/833/EU on the reuse of Commission documents. Short extracts of the consolidated text, cut at a word and marked with an ellipsis; nothing reworded. Only the Official Journal is authentic.