Rättskällor med officiella primärkällor

Utskrivet ·

Skip to main content
Skip to the answer

Agent · ai-act-2024-1689-15

AI Act artikel 15: Accuracy, robustness and cybersecurity

Structural tree: the article's own paragraphs, verbatim.

CELEX 32024R1689 · 2026-08-18 · Weight 88 · minimal-risk

PremiumHand written or high weight rule tree. Metered per call at the edge once metering is switched on, at the same address and with the same answer as today.

AI ActOfficial source

What this page is
Agent, AI Act artikel 15
Checked against the official source
2026-08-18Current
Responsible publisher
ExploreWorld Legal, editorial deskLiability position

Jurisdiction

The same agent, read through one country's lens.

Inputs

  • in_scopeThe article applies to the situationboolean
  • punktParagraph of the articleenum (1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9)

Rule tree

  1. If: alla(in_scope = true, punkt = 1)

    Paragraph 1 applies

    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.

    Paragraph 1

  2. If: alla(in_scope = true, punkt = 2)

    Paragraph 2 applies

    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.

    Paragraph 2

  3. If: alla(in_scope = true, punkt = 3)

    Paragraph 3 applies

    3. The levels of accuracy and the relevant accuracy metrics of high-risk AI systems shall be declared in the accompanying instructions of use.

    Paragraph 3

  4. If: alla(in_scope = true, punkt = 4)

    Paragraph 4 applies

    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.

    Paragraph 4

  5. If: alla(in_scope = true, punkt = 5)

    Paragraph 5 applies

    The robustness of high-risk AI systems may be achieved through technical redundancy solutions, which may include backup or fail-safe plans.

    Paragraph 5

  6. If: alla(in_scope = true, punkt = 6)

    Paragraph 6 applies

    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.

    Paragraph 6

  7. If: alla(in_scope = true, punkt = 7)

    Paragraph 7 applies

    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.

    Paragraph 7

  8. If: alla(in_scope = true, punkt = 8)

    Paragraph 8 applies

    The technical solutions aiming to ensure the cybersecurity of high-risk AI systems shall be appropriate to the relevant circumstances and the risks.

    Paragraph 8

  9. If: alla(in_scope = true, punkt = 9)

    Paragraph 9 applies

    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 mod…

    Paragraph 9

If no rule matches: The article is not stated to apply, or no paragraph is selected. The agent abstains rather than guesses.

The article text as read

  1. 11. 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.
  2. 22. 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.
  3. 33. The levels of accuracy and the relevant accuracy metrics of high-risk AI systems shall be declared in the accompanying instructions of use.
  4. 44. 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.
  5. 5The robustness of high-risk AI systems may be achieved through technical redundancy solutions, which may include backup or fail-safe plans.
  6. 6High-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.
  7. 75. High-risk AI systems shall be resilient against attempts by unauthorised third parties to alter their use, outputs or performance by exploiting system vulnerabilities.
  8. 8The technical solutions aiming to ensure the cybersecurity of high-risk AI systems shall be appropriate to the relevant circumstances and the risks.
  9. 9The 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.

Lineage

treatyTFEU art. 288 (förordning)
act32024R1689
chapterIII. High-risk AI systems
article15
paragraphs9
jurisdictionEuropean Union (EU)
supervisorEDPB — European Data Protection Board
national

Interface

callhttps://legal.exploreworldai.com/api/public/v1/agents/ai-act-2024-1689-15/run
methodGET
outputmatched, outcome, trace, missing, hash
Quota60 anrop per minut och adress, utan nyckel
stabilityRegelträdet versioneras. En ändring byter artefakthash, aldrig adress.

Hashes

textsha256:5784f7d01cb16ba34a822855d640fb55c8ed54d53a458d72099c6475814b25bb
scriptsha256:1011fef52a7bc5daa2c0b0a8f565e2f8d57edf3429e31695cea79958c9e31003
enginesha256:0a4bd50d21f8ec9be383fc091511008b76ad61909cbfb674eab56fe567fbd7a0
agentsha256:dfa5c377b9f202c1f30bd02d737dc2f3ad57b363d571b48425af0bf2d589922a
versionagent-engine-1+legal-2026-08-25 / dfa5c377b9f202c1

Artefacts

No legal advice. Deterministisk regeluppslagning. Ingen juridisk rådgivning, inget efterlevnadsbeslut, ingen bedömning av ett enskilt ärende.

Citation: 32024R1689 art. 15, Accuracy, robustness and cybersecurity. ExploreWorld Legal, https://legal.exploreworldai.com/agent/ai-act-2024-1689/artikel-15 (hämtad 2026-08-18, bevis sha256:855cd1db897845dd, bygge legal-2026-08-25).