High-Risk Annex I Regulated Product AI Obligations

Deadline and scoping page for AI embedded in Annex I regulated products such as medical devices, machinery, vehicles, toys, lifts, and radio equipment.

Legal status. The Digital Omnibus on AI was published in the Official Journal of the EU on July 24, 2026 as Regulation (EU) 2026/1744 and entered into force on July 27, 2026. The deferred dates below are now binding law.
Status last verified:

Changed by Omnibus Original: August 2, 2027

High-risk AI obligations for Annex I regulated products apply

Obligations for high-risk AI embedded in Annex I regulated products, including medical devices, machinery, vehicles, toys, lifts, and radio equipment, are moved from the original date by the Digital Omnibus on AI.

Changed by the Digital Omnibus on AI (Regulation (EU) 2026/1744), in force since July 27, 2026.

What the obligation actually requires

Annex I is for high-risk AI embedded in regulated products. The deadline data gives examples: medical devices, machinery, vehicles, toys, lifts, and radio equipment. The important distinction is that the AI system is tied to a product category that already has its own conformity and safety pathway. This is not the same as a standalone HR, credit, or education tool.

In practice, a flagged company should connect AI Act work with the product regulatory file. The AI feature should be described in the same operational language used by engineering, quality, and regulatory teams: what the product does, what the AI component controls or influences, what failures matter, and how the product will be documented for the relevant conformity process. If the AI is a safety component or is embedded deeply enough to affect the regulated product, it should not be reviewed in isolation by the software team.

Who usually triggers it

A medical-device company adding AI-assisted analysis to a device is a common example. A machinery manufacturer using AI in control or safety-related functions should review this path. A vehicle, toy, lift, or radio-equipment manufacturer that embeds AI into the product experience or product operation should also check whether the Annex I route is the better fit.

The common misunderstanding

The common mistake is confusing Annex I and Annex III. Annex III is about named standalone high-risk use cases. Annex I is about regulated products. A company can waste time if it treats an embedded product feature like an ordinary software workflow, or if it ignores the existing product conformity process and builds a separate AI compliance track that does not connect to the real technical file.

What to do next

Start with the product list, not the AI vendor list. Identify every regulated product that contains AI or depends on an AI component. For each product, ask whether the AI affects safety, conformity, or user-facing operation. Then bring product regulatory, quality, engineering, and legal owners into one review. The timeline above shows the current and original dates from the deadline data; use that runway to align documentation rather than creating a last-minute AI-only appendix.

Use the deadline checker View full timeline