The EU AI Regulation (EU AI Act) is now progressively in force. This is no longer a distant prospect. If your company uses or develops AI systems, even through third-party tools, you have obligations. Here's the plain-language guide to understanding where you stand.
The EU AI Act classifies AI systems into 4 risk levels. This determines all your obligations. The big mistake SMEs make is thinking the text doesn't apply to them.
| Level | Examples | Obligations | Max fine |
|---|---|---|---|
| Unacceptable | Social scoring, manipulation, real-time facial recognition | Total ban | , |
| High risk | HR AI, credit scoring, medical and educational systems | Testing, logs, human oversight, technical documentation | €30M or 6% turnover |
| Limited risk | Chatbots, content generators | Transparency: disclose AI interaction | €15M or 3% turnover |
| Minimal risk | Spam filters, non-critical recommendations | No specific obligations | , |
If you deploy a third-party AI system (HR SaaS, credit scoring tool) in a high-risk context, you have obligations as a deployer. Responsibility doesn't lie solely with the vendor.
We run EU AI Act compliance audits tailored to your sector and size. We tell you exactly where you stand and what to do first. First audit free, no commitment.
In 60 minutes, we analyse your situation and deliver a clear report with priority actions. Free, no commitment.
Book my free audit