The AI Act has two routes into high-risk. One is Annex III, covering use cases such as recruitment. The other — less well known but central to transport — is Annex I, which applies to AI as a safety component in products already regulated by EU sectoral product legislation. For vehicles, rail and other transport equipment, it is often Annex I that governs.
Two routes to high-risk
An AI system can be classified as high-risk in two ways:
- Annex III — the system is used within a listed area, e.g. employment (recruitment, staffing).
- Annex I — the system is a safety component of, or is itself, a product covered by the Union legislation listed in Annex I, and the product requires third-party conformity assessment.
For transport, route 2 is common: AI in braking systems, driver assistance, signalling or autonomous control is typically a safety component under existing product legislation.
What Annex I means in practice
Annex I links the AI Act to the sectoral legislation that already applies to the product — for example, type-approval of motor vehicles or rules for railway equipment. The aim is to avoid duplicate, parallel systems: the requirements for the AI system should as far as possible be integrated into the conformity assessment already carried out for the product.
Article 43(3): sectoral conformity assessment
Article 43(3) governs how conformity is assessed when a high-risk AI system is covered by such sectoral legislation in Annex I. The main principle is that the AI Act's requirements are assessed within the existing sectoral procedure, by the body already responsible for assessing the product, rather than in a separate AI process. This means the AI requirements and the product requirements are handled in the same procedure.
For you as a transport company, it is crucial to know which track your system belongs to: an Annex I product is handled together with the product approval, while an Annex III system (such as recruitment) follows the ordinary high-risk procedure and the deployer obligations in Article 26.
Dates
- 2 August 2026 — general application of the AI Act, including Annex III obligations.
- For high-risk AI tied to products under Annex I, special and often later transition periods apply — check the sectoral legislation governing your specific product.
The Digital Omnibus proposal to postpone parts of the high-risk rules to 2 December 2027 was approved by the European Parliament on 16 June 2026 but is not in force.
Penalties
Article 99: up to EUR 15m or 3% of global turnover for breaching the high-risk requirements, and EUR 7.5m or 1% for incorrect information. Prohibited practices (Article 5) can reach up to EUR 35m or 7%.
The right track, the right evidence
Determining whether your AI system runs via Annex I or Annex III is the first step — it shapes the entire compliance path. PowerQuant helps transport companies classify the system correctly and build the compliance evidence (M1/M2/M3) that matches the right track.
This page is general information about the AI Act and does not constitute legal advice. Classification of an individual system should be done with legal and technical expertise.