EU AI Act for Small Businesses
If your business runs an AI chatbot, an AI voice assistant, or you publish anything an AI produced (text, images, audio or video), the transparency rules in Article 50 of the EU AI Act start applying to you on 2 August 2026. That holds whatever your size or sector. The good news is that the basics do not require a lawyer. You have to tell people when they are dealing with AI, label the content your AI creates, and keep a short record of what you did. This page walks through it in plain English for small and medium businesses in Cyprus and Greece.
Does the EU AI Act apply to my small business?
In most cases, yes. If you use generative AI anywhere a customer can see it, you are in scope. Article 50 is not limited to systems labelled “high risk”. It can apply to many organisations that run a customer-facing AI chatbot, use AI to generate content, or publish synthetic media, from a marketing shop to a law firm to an online store, and there is no blanket carve-out for small companies. One common-sense limit applies to chatbots: you do not have to announce the AI when it is already obvious to a reasonably informed person that they are dealing with one.
What are the four transparency obligations?
- Chatbots and AI assistants. People must be told clearly that they are talking to AI rather than a person, unless it is already obvious.
- AI-generated content. Synthetic text, images, audio and video have to be marked in a machine-readable way and be detectable as AI-made.
- Emotion recognition and biometrics. You have to inform people when a system like that processes their biometric data.
- Deepfakes. Media that looks like real people, places or events but was generated or altered by AI has to be disclosed as artificial.
When is the deadline, and is there really an extension?
The main date is 2 August 2026. There is a lot of noise about an “extension”, so here is the accurate version. Under the Digital Omnibus, the content-marking rule in Article 50(2) only starts on 2 December 2026 for AI systems that were already on the market before 2 August 2026, which gives a four-month transition. Anything you launch on or after 2 August 2026 has to comply from day one. For practical purposes, treat 2 August 2026 as your date.
What should an SMB actually do before then?
- List every chatbot, AI assistant, and place where you publish AI-generated content.
- Add a clear notice to each chatbot, something like “You are chatting with an AI assistant”.
- Turn on machine-readable marking for AI content wherever your tools support it, and label it for readers too.
- Disclose any synthetic or deepfake media visibly, right where it appears.
- Write a one-page internal note recording what you changed. That note is your evidence of good faith.
- Make sure the team understands the tools they use. AI literacy has been a legal expectation since February 2025 under Article 4.
Want a quick read on where your chatbot stands? Our free two-minute check gives you a plain verdict on your Article 50 exposure.
What are the penalties?
The AI Act allows for meaningful administrative fines, but for most smaller businesses the realistic risk is reputational damage and unwanted attention rather than a headline penalty. Both are avoidable with the steps above. Following the Commission’s voluntary Code of Practice on AI-generated content can also count in your favour if a question ever comes up.
How Dabster Labs can help
We build and run AI chatbots and automations for SMBs across Cyprus and Greece, with the disclosure, labelling and record-keeping built in from the start and GDPR handled properly. If you would rather not work through this on your own, book a free automation audit and we will map your AI systems to a concrete plan.
This guide is educational and is not legal advice. For a binding assessment, speak to a qualified professional.
