Annex III point 2 lists AI systems intended to be used as safety components in the management and operation of critical digital infrastructure, road traffic, and the supply of water, gas, heating and electricity as high-risk use cases, subject to the Article 6 conditions. AI systems that are safety components of regulated products in those sectors can also follow the Annex I route. Registration for Annex III point 2 systems is at national level under Article 49(5), not in the EU database. The Annex III rules apply from 2 December 2027 and the Annex I rules from 2 August 2028, as amended by Regulation (EU) 2026/1744. This guide does not decide whether any infrastructure system is a safety component or is high-risk, and it does not give safety engineering advice.
What is a safety component?
A safety component is a component of a product or system that fulfils a safety function, or whose failure or malfunction endangers the health and safety of persons or property. The definition in the Act should be checked in the consolidated text, since it turns on the function the component performs and the consequence of its failure, not on whether the component is labelled safety-critical by its supplier. In infrastructure, a safety component can be a control system that manages pressure, flow, voltage or traffic, a protection system that detects a fault and isolates it, or a monitoring system whose failure would leave a hazard undetected. A system that only reports, forecasts or optimises, without fulfilling a safety function, is assessed differently, and the distinction must be recorded from the system's function and failure modes. The safety component question should be assessed with sector safety expertise, since it requires an understanding of the infrastructure, its hazards and the consequences of failure.
Which infrastructure is listed in Annex III point 2?
Annex III point 2 lists AI systems intended to be used as safety components in the management and operation of critical digital infrastructure, road traffic, and the supply of water, gas, heating and electricity. Each area must be tested on its own terms: critical digital infrastructure covers the digital systems on which other infrastructure depends, road traffic covers the management and operation of traffic systems, and the supply areas cover the networks that deliver water, gas, heating and electricity. A system that manages a network's safety functions is within the listing's language, while a system that only handles billing, customer information or long-term planning for the same network may not be. The record should describe the infrastructure, the safety function, the management or operation role, and the consequence of failure, and should cite the Annex point for each function. A system that serves more than one area should be assessed for each area it serves.
For the broader regulatory framework, see finance, insurance and critical infrastructure.
How do the Annex III and Annex I routes differ here?
The Annex III route covers standalone systems used as safety components in the management and operation of critical infrastructure, with the Chapter III requirements applying from 2 December 2027 and registration at national level. The Annex I route covers AI systems that are safety components of products, or are themselves products, covered by the Union harmonisation legislation in Annex I, where third-party conformity assessment is required, with the Chapter III requirements applying from 2 August 2028 and conformity through the sector legislation. An embedded control unit in a regulated product follows the Annex I route, while an operational management system for a network follows the Annex III route, and a system that is both embedded and operational can raise both questions. The route record should be made before a deadline or a conformity plan is assigned, since the wrong route produces the wrong deadline, the wrong registration treatment and the wrong conformity procedure.
Why is registration national for point 2 systems?
High-risk AI systems referred to in Annex III point 2 are registered at national level under Article 49(5), rather than in the EU database. That national route reflects the national supervision and security sensitivity of critical infrastructure, and it means the provider should confirm the national registration process, the authority concerned and the information required in each Member State where the system is placed on the market or put into service. A provider operating in more than one Member State should record the national route separately for each State, since a registration in one State does not automatically satisfy the position in another. The national registration record should be kept with the system's compliance file, with the authority, the date and the information submitted recorded, in the same way an EU database entry would be. The absence of an EU database entry for a point 2 system is therefore expected, and should not be treated as a missing registration.
How should AI Act and sector safety work be coordinated?
AI Act work and sector safety work should be coordinated in a single safety and compliance file per system version, carrying the safety function, the hazard analysis, the risk management, the data governance, the technical documentation, the testing and validation evidence, the oversight design, the incident and resilience arrangements, and the registration record. Sector safety regulation, including resilience, incident reporting and operational security duties, can impose requirements that overlap with the AI Act's requirements on accuracy, robustness, cybersecurity and post-market monitoring, and the file should show how each requirement is met and where the regimes differ. Changes to the system, its data, its operating environment or its safety function should trigger a joint review, since a change that is minor under one regime can be significant under the other. The file should be available to the market surveillance authority, the sector regulator and the notified body where applicable, in the form each requires, without describing the system differently in each forum.
Continue with energy, water and digital infrastructure.
What does this guide not decide?
This guide does not decide whether any infrastructure system is a safety component, is high-risk, or follows the Annex I or Annex III route, and it does not give safety engineering advice on the design or operation of critical infrastructure. Those questions require the system's function, failure modes, sector legislation and safety analysis to be examined together, with sector safety expertise and qualified advice. The guide also does not describe any Member State's national registration procedure, which must be confirmed with the authority concerned. Where a critical infrastructure system is to be deployed, confirm the route, registration, conformity and sector safety position with the competent authorities before the system is put into service.
Frequently asked questions
Are critical infrastructure AI systems in the EU database?
No. Annex III point 2 systems are registered at national level under Article 49(5). The national process, authority and information required must be confirmed in each Member State concerned.
What makes a component a safety component?
A component that fulfils a safety function, or whose failure or malfunction endangers health, safety or property, as defined in the Act. The function and the consequence of failure must be assessed with sector safety expertise.
Which deadline applies to infrastructure systems?
It depends on the route. Annex III systems apply from 2 December 2027, Annex I product systems from 2 August 2028. The route must be identified before a deadline is assigned.
Continue through this AI Act cluster
This article is an informational overview, not a legal determination. Check the current legal text and official implementation guidance for your system and role.
Consolidated Regulation (EU) 2024/1689, as amended ↗AI Act Service Desk implementation timeline ↗