What the EU AI Act means for Swiss SMEs
The regulation applies in the EU — and yet it affects many Swiss businesses. What you need to know without being a lawyer.
What it is about
Regulation (EU) 2024/1689, the AI Act, has been in force since August 2024 and takes effect in stages. It sorts AI applications by risk: prohibited, high risk, limited risk, minimal risk. For most SME uses — chat assistants, text suggestions, image generation — the “limited risk” tier applies, and there it is mainly about transparency.
Transparency means: anyone talking to an AI must be able to tell. AI-generated images, voices and videos must be labelled as such. These duties under Article 50 apply from 2 August 2026.
Why it concerns Swiss businesses
The regulation is not tied to where a company is based but to where the AI’s output is used. A Swiss hairdresser with clients from Konstanz, a web shop delivering to Germany, a cleaning company near the border — as soon as AI output reaches the EU, the regulation is in play.
In Switzerland itself, the revised Data Protection Act (FADP) has applied since September 2023. In 2025 the Federal Council announced it would adopt the Council of Europe’s AI Convention and regulate sector by sector — there is currently no Swiss AI law modelled on the EU’s.
What it means in practice
For a business it comes down to three things: label AI images and AI voices, introduce AI assistants as such, and never hand personal data unprotected to third-party models. All three are solvable — and all three are our standard, not an add-on. How we handle customer data is explained in the article on masking.
This text is a general overview, not legal advice.