Skip to content

AI Act4 November 2026

The AI Act: what fines can an SME get and how to avoid them

Xora Intelligence4 min read
Two partners in an SME read the regulation with marked pages on a rainy day

An SME that breaches the EU AI Act (Reglamento europeo de IA) faces a fine of up to 7%, 3% or 1% of its worldwide annual turnover, or €35, €15 or €7.5 million, depending on the infringement. For SMEs the lower of the two figures applies. These are caps, not fixed fines.

We are Xora Intelligence, an AI consultancy in Los Barrios (Cádiz). This is not legal advice: it is a map of what to look at first. If you want all the obligations and the full calendar, you have them in our guide to the Regulation for SMEs.

How much can an SME end up paying?

Article 99 of Regulation (EU) 2024/1689 sets three levels. We have worked out the cap for an example SME with a turnover of €400,000 a year:

Infringement General cap Cap for that SME
Using a prohibited practice (Art. 5) €35M or 7% €28,000
Failing on transparency (Art. 50) or high-risk obligations €15M or 3% €12,000
Giving false or incomplete information to the authority €7.5M or 1% €4,000

That is the maximum, not what you would be given. The article itself says to weigh the seriousness, the duration, how many people were affected, whether there was intent or negligence, whether you cooperated and what you did to reduce the harm.

What conduct is punished in an ordinary SME?

What is most likely to affect you depends on how you use AI, not on whether you use it. These are the three cases to check:

  • A chatbot that doesn't say it is a chatbot. If a person is talking to an AI, they have to know, unless it is obvious. This applies from 2 August 2026. It is fixed with one sentence in the first message.
  • Fake content made with AI. If you publish a deepfake, an image, audio or video that looks real, you have to state that it is artificial.
  • Prohibited practices. For example, using AI to infer your employees' emotions at work, except for medical or safety reasons. These bans have applied since February 2025.

Using AI to filter CVs or assess a person's creditworthiness is high risk. Those obligations were postponed to 2 December 2027, but it is worth knowing before you buy the tool.

What about the obligation to train the team (Art. 4)? Article 99 does not assign it a fine of its own. Even so, having it done is the first thing you will want to show if someone asks: we explain it in how to train your team in AI.

Who can fine you today in Spain?

There is news here that many pages have not picked up yet. The Spanish organic law bill, which was going to set out the authorities and the penalty procedure, has appeared as lapsed since 6 October 2026 on the Congress website. It fell with the dissolution of the Cortes before the elections of 29 November. If the next Government wants that law, it will have to start again.

That does not mean you can forget about it:

  • The Regulation is European and applies directly. Your obligations do not depend on that law.
  • AESIA still exists (Spain's AI supervision agency), and the IA360 Plan names it as the authority that will carry out market surveillance.
  • What remains unspecified is how a penalty would be processed in Spain and the exact split of powers.
  • The GDPR and the LSSI (Spain's information society services law) remain fully in force. If your AI processes customer data, that risk exists today, with its own authority and its own fines. We look at it in AI and the GDPR with customer data.

How do you reduce the risk this week?

Five tasks, without buying anything:

  1. Make a list of the AI tools used in the company, including the ones nobody approved.
  2. Mark each use: does it talk to customers? Does it publish content? Does it decide about people?
  3. Add the notice to your chatbots and voice assistants.
  4. Write a page of rules and give a short training session. Keep the date and who attended. You have an AI use policy template ready to fill in.
  5. Keep the supplier's documentation before using AI in recruitment or credit.

What Xora Intelligence does in this case

In a company audit we review which AI you use, what data it touches and where you have risk, and we leave you a prioritised list. For the final legal assessment of your case, we will recommend that you involve a lawyer. You can write to us through contact.

Frequently asked questions

Can I be fined for using ChatGPT at work? Using it is not penalised in itself. The risk lies in what you use it for: decisions about people, fake content without a notice, or customer data entered without a legal basis.

Is the fine for an SME always a percentage? The lower of the fixed amount and the percentage applies. For almost any SME, the lower one will be the percentage of its worldwide annual turnover.

Do these fines already apply? The fines regime has applied since 2 August 2025, and most of the obligations that affect an SME using AI are already in force. In Spain the procedure remains to be specified after the bill lapsed.

Am I responsible if the fault is my AI supplier's? It depends. The supplier has its obligations and you, as the person using the tool, have yours. Telling people on your website that a chatbot is an AI, for example, is something you should check yourself.

Do a micro-business or a self-employed person have fewer obligations? The Regulation provides support measures for SMEs and lower caps, but it does not exempt them from the prohibitions or from transparency.

Found it useful? Share it with someone who needs it

Xora IntelligenceAI consultancy and custom software in Los Barrios (Cádiz). We write about what we build for businesses in the province. If you have a question about your case, write to us.

Keep readingRelated articles

The whole blog

Bay of Gibraltar · photo by Cabeza2000 (CC BY-SA 4.0)

Shall we talkabout your business?

Tell us which tasks take up most of your time. We will tell you whether they can be automated and how we would do it. If we can't help, we will tell you that too.