Article 15 for Autonomous Systems: Accuracy, Robustness and Cybersecurity

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.

Article 15 binds high-risk AI systems only. The rows below show which autonomous systems it reaches and which it does not.

• Autonomous systems is not a category in the Act

  The Act has no category called autonomous systems; its definition of an AI system refers to operation with varying levels of autonomy (Article 3(1)). This site uses the word for AI in vehicles, drones, robots and machinery, and for AI that helps manage road traffic. Such AI can be high-risk through three routes, and the route decides which high-risk rules apply: Annex I Section B (Article 2(2) limits the rules to a few articles), Annex I Section A (the Chapter III requirements from 2 August 2028, where the system is high-risk under Article 6(1)) and Annex III point 2 (the Chapter III requirements from 2 December 2027). Other rules do not depend on the route: Articles 4 and 5, Article 50, Article 53 for model providers and the other points of Annex III. Article 2(2) restricts only AI systems that are high-risk under Article 6(1) in Section B products. Find your high-risk route first.

• Vehicles, aircraft, rail, marine equipment and machinery: Annex I Section B

  Section B of Annex I lists, among others, motor vehicles (Regulations (EU) 2018/858 and 2019/2144), two- or three-wheel vehicles (Regulation (EU) No 168/2013), agricultural and forestry vehicles (Regulation (EU) No 167/2013), the rail system (Directive (EU) 2016/797), marine equipment (Directive 2014/90/EU), unmanned aircraft and their engines, propellers, parts and equipment to control them remotely, in so far as Regulation (EU) 2018/1139 covers them, and machinery (Regulation (EU) 2023/1230, moved from Section A to Section B by Regulation (EU) 2026/1744). For an AI system that is high-risk under Article 6(1) and relates to such a product, only Article 6(1), Article 60a and Articles 102 to 112 apply (Article 2(2)); Articles 57, 58 and 59 apply only in so far as the requirements for high-risk AI systems have been integrated in that Union harmonisation legislation. Articles 9 to 15, 17 and 26 are not among them. Instead, Articles 102 to 109 add to eight of these laws a rule that the Commission takes the requirements of Chapter III, Section 2 into account when it adopts the acts they name for AI systems. For machinery, Article 3(1) of Regulation (EU) 2026/1744 requires the Commission to adopt delegated acts that add health and safety requirements for such AI systems, reflecting Chapter III, Section 2 and Articles 17, 19, 72 and 73; those acts are to apply by 2 August 2028.

• Section A products and Annex III point 2: Chapter III applies

  Two hooks bring the Chapter III requirements into play. First, an AI system is high-risk under Article 6(1) if it is intended to be used as a safety component of a product covered by Annex I legislation, or is itself such a product, and that product must undergo a third-party conformity assessment under that legislation. Section A (items 2 to 12) covers, for example, lifts, cableway installations, recreational craft and medical devices; Chapter III applies to such systems from 2 August 2028 (Article 113, point (c)(ii)). A safety component is a component that fulfils a safety function or whose failure or malfunctioning endangers the health and safety of persons or property (Article 3(14)); AI used solely for non-safety related aspects such as performance optimisation, automation or convenience does not qualify, but AI whose failure would endanger health and safety does (Article 6(1a) and (1b)). A product that needs a third-party assessment solely because of risks other than health and safety, such as radio spectrum use or electromagnetic interference that does not affect health and safety, does not meet that condition (Article 6(1c)); this matters, for example, for radio equipment (item 6). Second, Annex III point 2 covers AI systems intended to be used as safety components in the management and operation of critical digital infrastructure, road traffic, or the supply of water, gas, heating or electricity; Chapter III applies to them from 2 December 2027 (Article 113, point (c)(i)). On its wording, point 2 names the management and operation of road traffic, not vehicles.

• Provider and deployer

  The provider develops the AI system, or has it developed, and places it on the market or puts it into service under its own name or trademark (Article 3(3)). For a high-risk AI system that is a safety component of a Section A product, the product manufacturer is the provider if the system is placed on the market with the product under the manufacturer’s name or trademark, or put into service under that name or trademark after the product has been placed on the market (Article 25(3)). Article 15 is a design duty of the provider. A fleet, plant or traffic operator that only uses the system is a deployer (Article 3(4)) and has no Article 15 duty; a deployer becomes a provider in the cases of Article 25(1).

• Dates

  Annex III point 2 systems must meet Chapter III from 2 December 2027, and Section A systems from 2 August 2028 (Article 113, point (c)). For Section B systems, no date applies to Articles 9 to 15, 17 and 26, because those articles do not apply to them; Article 113, point (c)(ii), applies Chapter III, Sections 1 to 3, which include Article 6, to Annex I systems from 2 August 2028, and Articles 102 to 110 apply since 27 July 2026 (Article 113, points (c)(ii) and (d)). A system placed on the market or put into service before the date of application of Chapter III 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)).

• Fines

  Breaches of provider obligations under Article 16 and of deployer obligations under Article 26 can be fined up to EUR 15 million or, for an undertaking, up to 3% of total worldwide annual turnover in the preceding financial year, whichever is higher (Article 99(4)). For SMEs, including start-ups, each fine is capped at whichever is lower (Article 99(6)). The same cap applies to small mid-cap enterprises (Article 3(14b), Article 99(6a)). Each Member State lays down rules on to what extent administrative fines may be imposed on public authorities and bodies established in that Member State (Article 99(8)). Article 99 is not among the provisions that Article 2(2) lists as applying to AI systems that are high-risk under Article 6(1) in Section B products.

• Limits of the Act’s scope

  The Act applies to providers that place AI systems on the market or put them into service in the Union, or place general-purpose AI models on the market in the Union, wherever they are established (Article 2(1), point (a)), and to providers and deployers in a third country where the output of the AI system is used in the Union (point (c)). It does not apply to areas outside the scope of Union law and does not affect Member States’ competences on national security, nor to AI systems used exclusively for military, defence or national security purposes (Article 2(3)). The Act does not apply 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)). It does not apply to research, testing or development before a system or model 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 as an AI system that falls under Article 5 or 50 (Article 2(12)).

• No accuracy values for autonomous systems

  Article 15 names no accuracy metric, minimum value, test method or standard for any sector, and none for autonomous systems. The provider declares the levels and relevant metrics (Article 15(3)). Harmonised standards whose references are published in the Official Journal give a presumption of conformity in so far as they cover the requirements (Article 40(1)); a published cybersecurity certificate under Regulation (EU) 2019/881 gives one for the cybersecurity requirements in so far as it covers them (Article 42(2)). Systems that fall within the scope of Regulation (EU) 2024/2847 and meet the conditions in Article 12(1) of that Regulation are deemed to comply with the cybersecurity requirements of Article 15 (Article 42(3)). The Commission is to develop guidelines on the application of the requirements and obligations in Articles 8 to 15 and Articles 25 and 26 (Article 96(1), point (a)).

• Sector standards and type-approval rules are not Act duties

  The Act names no ISO, UNECE or other sector standard. Which accuracy, robustness or cybersecurity rules a Section B product must meet is decided by its sector law, not by Article 15; the Act is without prejudice to other Union legal acts on product safety (Article 2(9)). The Commission is to develop guidelines on the relationship of the Act with the Union harmonisation legislation listed in Annex I and with other relevant Union law (Article 96(1), point (e)).

• Section A: requirements may be limited (Article 2(13))

  For high-risk AI systems referred to in Article 6(1), the specific requirements or obligations in Articles 9 to 15 and 17 to 25 may be limited, where and to the extent that Union harmonisation legislation listed in Section A of Annex I lays down requirements or obligations providing an equivalent or higher level of protection and the limitation does not reduce the overall level of protection under the Act (Article 2(13)). The Commission is to adopt delegated acts specifying the systems concerned, the requirements that may be limited, the conditions and the scope by 2 August 2027. On a reading of Article 2(13), the Act itself does not say which requirements of a given Section A product are limited; the delegated acts are to do so; ask counsel.

• Example: declared metrics for a traffic-management component

  Example, not an Act requirement: the provider of a safety component for road-traffic management could declare which accuracy metrics it measured and under which road, sensor and weather conditions it measured them.

• Example: documenting the fail-safe behaviour

  Example, not an Act requirement: a provider could document what the system does when a sensor fails or its input is implausible, for example by handing control to non-AI logic, as one way to describe the fail-safe plans mentioned in Article 15(4).

• EU AI Act Article 15: Accuracy, Robustness and Cybersecurity

  Article 15 for all sectors: the requirements, dates and fines.

• EU AI Act Compliance for Autonomous Systems

  The autonomous systems overview: which Act hooks apply to vehicles, aircraft, machinery and traffic AI.

• Article 17 for Autonomous Systems: Quality Management System

  Autonomous systems and Article 17: the quality management system and who is the provider.

• Annex I High-Risk AI Systems: The Product-Safety Route

  Annex I: which AI systems in regulated products are high-risk, with Section A and Section B.