My supplier says the product has AI. Is it a high-risk AI system?
Usually not. Article 6(1) requires two conditions at once: the AI must be a safety component of, or itself be, a product covered by the Annex I legislation, and that product must be required to undergo third-party conformity assessment. Many Annex I products self-certify — and where the product self-certifies, the AI inside it is not high-risk by that route.
This area changes quickly. The explanation below is written to stay true as the numbers move, and deliberately avoids quoting a rate or threshold that would age badly. Check the official sources at the foot of this page, and confirm with a licensed professional before you rely on it. Last checked 1 September 2026.
This is worth getting right rather than assuming, in both directions. Answered “yes” when the answer is no, a buyer commissions a conformity programme nobody asked for. Answered “no” when the answer is yes, the goods cannot lawfully be placed on the market at all.
The test has two limbs, and most summaries quote one
Under Article 6(1), an AI system is high-risk by the product route only where both of the following are true:
- the AI system is a safety component of a product, or is itself a product, covered by the EU harmonisation legislation listed in Annex I; and
- that product is required to undergo third-party conformity assessment under that same legislation before it may be placed on the market.
The second limb is the one that gets dropped, and it is the one that decides most cases. Many Annex I products self-certify: machinery outside the specifically listed categories, Class I medical devices, and toys where the harmonised standards are applied in full. Where the product self-certifies, the first limb is met and the second is not — so the AI in it is not high-risk by this route.
The Annex I product legislation
| Annex I | What it covers |
|---|---|
| Section A | Machinery · toys · recreational craft and personal watercraft · lifts and safety components · equipment for potentially explosive atmospheres · radio equipment · pressure equipment · cableway installations · personal protective equipment · appliances burning gaseous fuels · medical devices · in vitro diagnostic medical devices |
| Section B | Civil aviation security · two- and three-wheel vehicles · agricultural and forestry vehicles · marine equipment · rail interoperability · motor vehicles · unmanned aircraft. Here the AI Act’s requirements are folded into the sectoral regimes rather than applied directly |
Settle the product question before the AI question. Does this product need a notified body under its own legislation? That is an ordinary compliance question you can answer from the harmonised standards and your existing test reports, and it disposes of the AI classification for most consumer goods. Whether the certificates you were sent are genuine, current and actually cover your model is a separate exercise — and usually the more productive one.
The other route: Annex III
A system can also be high-risk on its own, without any product wrapper, if it falls into one of the eight Annex III areas: biometrics, critical infrastructure, education and vocational training, employment and worker management, access to essential private and public services, law enforcement, migration and border control, and the administration of justice. Almost nothing arriving in a container from a factory is in that list. These requirements apply from 2 December 2027.
And is it even an AI system?
Often it is not. The word on the Chinese datasheet is usually 智能, which translates as “smart” at least as often as “intelligent” — a 智能 kettle has a thermostat. Article 3(1) turns on inference: a system that infers from its input how to generate predictions, content, recommendations or decisions. The Commission’s guidelines of 6 February 2025 place simple deterministic rules, basic statistical estimators and plain data visualisation outside that definition. Ask the factory whether the product changes what it does based on data it collects, and where the model runs.
The dates, the roles and the questions to put to a supplier are set out in the EU AI Act and Chinese-made products.
Want this done rather than explained?
Is that CE certificate real, current, and for your model? — US$120, 1–2 working days.
Sources
- AI Act Article 6 — classification rules for high-risk AI systems
- AI Act Annex I — the Union harmonisation legislation that routes AI into the high-risk category
- European Commission — guidelines on the definition of an AI system, 6 February 2025
- Regulation (EU) 2026/1744 — the Digital Omnibus on AI, which moved the high-risk dates
All sources checked 1 September 2026. This page is general information, not legal, tax or customs advice. Requirements vary by product, market and circumstance — confirm your own position before acting.