# EU AI Act for Hardware: Which Products Are Actually Affected

The EU AI Act is not just a software regulation. Any hardware that hosts or deploys an AI system — from smart cameras and diagnostic devices to autonomous robots and access control systems — is in scope. The classification determines your conformity assessment route, your documentation burden, and whether you need a third-party notified body. Here's how to map your product to the right risk class.


## Transcript

### Hardware that runs AI is in scope

Most founders assume the EU AI Act is a software problem. It isn't. It regulates AI systems — any hardware running inference or generating outputs that influence real decisions. If yours does that, you're in scope, and your risk class decides what comes next.

### Four risk classes and what they mean

The Act creates four risk tiers. Unacceptable risk is banned outright. High-risk covers AI in regulated products, plus uses like biometric ID — it demands conformity assessment and EU database registration. Limited-risk needs lighter transparency; minimal-risk needs nothing extra.

### How to assess your product

Start with Annex I and Annex III. Annex I lists regulated sectors where embedded AI turns high-risk; Annex III lists eight standalone high-risk use cases. If you fall into either, document your assessment — surveillance authorities will ask. Start now, not at CE marking time.

Tags: EU AI Act, hardware compliance, CE marking, high-risk AI, AI regulation