Skip to content

AI and GDPR21 July 2026

AI and the GDPR: the basics with your customers' data

Xora Intelligence12 min read
The manager of a clinic locks the cabinet holding customer records

Yes, you can use artificial intelligence with your customers' data, but the GDPR (RGPD) still applies in full. The basics come down to four things: having a legal reason to process each piece of data, telling people there is an automated system involved, being able to deal with their rights and not letting the machine decide on its own anything that really affects them. This article is the short version, written for anyone about to put in a chatbot or an automation. The practical detail (contract with the supplier, transfers outside the EU, impact assessment, health data) is in the complete guide to AI and the GDPR for SMEs.

Antonio runs a home renovation company in San Roque. This month he has decided to put an assistant on his website: something that answers at night, collects the details of anyone asking for a quote and books the visit for the next morning. He sits down to set it up and, just before pressing "publish", he stops. On his computer he has the name, phone number and address of everyone who has asked him for a quote in the last two years. And now he asks himself what anyone asks before connecting an AI to their business: can I put all this into an automated system? Do I have to tell anyone? Can I be fined?

Those are the right questions. The good news is that the law does not forbid using artificial intelligence with customer data. What it does is set conditions, and those conditions are quite reasonable when you read them slowly. This article explains them in plain English, thinking of an SME that is going to install a chatbot, an automation or any AI tool that touches personal data. It does not replace a lawyer or a data protection adviser, but it helps you know which questions to ask and avoid a nasty surprise.

Do I need permission to put AI to work on my customers' data?

Before talking about AI, we need to talk about an idea that is the foundation of the whole General Data Protection Regulation (Reglamento General de Protección de Datos, RGPD): to process personal data you need a reason the law recognises. That is called a lawful basis, and it is not a formality, it is the starting point. If you do not have one, you cannot process the data, whether you use AI or not.

The most common bases in a small business are four. The customer's consent, when they say yes clearly. Performance of a contract, when you process their data because you are providing a service they asked you for. Compliance with a legal obligation, such as keeping invoices. And legitimate interest, which allows certain reasonable processing as long as it does not override the person's rights and you have done the work of weighing it up.

Here comes the nuance many people miss. Having a customer's data for one reason does not mean you can use it for anything else. If Antonio collected a customer's phone number to carry out a renovation, that number is there because of the contract. Using it to answer questions and coordinate the visit falls within that purpose. Taking it to train an AI model that he will then sell or use for something else is a different purpose, and that needs its own justification.

The guide from the Spanish data protection authority (AEPD) on GDPR compliance for processing that involves artificial intelligence makes it clear: adding AI to a data processing operation removes no obligation under the GDPR, it keeps them all and also requires you to think about the risks specific to the technology. Put another way: AI is not an excuse to skip the rules, it is one more reason to keep them in mind.

In practice, for an SME this comes down to two honest questions. Why do I have this data? And does what I am going to do with the AI fit that reason? If the answer is yes, you are on track. If the AI leads you to use the data for something new, that is where you stop and think about the basis.

Do I have to say that I use artificial intelligence?

Yes. The GDPR requires you to inform people about what you do with their data, and that duty does not disappear because an AI sits in the middle. On the contrary: the less obvious it is to the customer what happens to their information, the more it matters to explain it well.

Informing is not a paragraph hidden in small print. It is explaining, in language people understand, who processes the data, for what, on what lawful basis, how long it is kept and what rights the person has. The natural place for that is your privacy policy. If you are going to bring AI into customer service or your processes, that document has to reflect it. It will not do for it to describe a business that is no longer yours.

Here it is worth being honest on one point: you do not need to invent promises or grandiose clauses. You do not have to promise where your servers are or sign commitments you cannot keep. What you do have to do is say what really happens with the data and refer the user to your privacy policy, where the details are kept up to date. Transparency done well is boring and clear, not spectacular.

The AEPD guide insists on the same line: when processing involves AI, the information given to the data subject has to be enough for them to understand what it implies, above all if the system is going to influence decisions that affect them. A customer has the right to know that, when they write to your website at eleven at night, the one answering is an automated system and not a person. In fact, that specific obligation (saying you are talking to a machine) is no longer just good practice: it has been law since August 2026, and a section below covers it.

What rights does the customer keep when you use AI?

All the ones they had before. Putting AI into your processes does not cut a single right of the person whose data you process. They stay intact, and you still have to honour them.

The customer can ask you for access to the data you hold about them. They can ask you to rectify whatever is wrong. They can ask you to erase their data when there is no longer a reason to keep it. They can object to certain processing and ask you to restrict its use while something is clarified. And they can take their data elsewhere in a usable format, which is called portability.

What matters for an SME is that these rights must be able to be exercised in reality, not just in theory. If you set up an AI system that accumulates conversations, histories and contact details, you have to be able to find everything about one specific person and act on it when they ask: show it to them, correct it or delete it. A system that stores data but from which you cannot cleanly extract it leaves you in a bad position the day someone exercises their right. That is why it pays to think about this when designing the tool, not when the first request arrives.

Can artificial intelligence decide alone about a customer?

This is the question where the law is most specific, and it deserves a careful read. Article 22 of the GDPR gives everyone the right not to be subject to a decision based solely on automated processing that produces legal effects concerning them or similarly significantly affects them. You can read the text of Article 22 in the Regulation itself.

In plain terms: if a machine, entirely on its own and with no person reviewing, makes a decision that seriously changes someone's life (refuses them credit, rejects them for a job, excludes them from a service), the law limits that. It is not that it is always forbidden: the article itself provides exceptions, such as when it is necessary for a contract, when the person has given explicit consent or when a rule authorises it. But even in those cases safeguards must be given, and one of the main ones is that the person can request human intervention, express their point of view and contest the decision.

The word that carries all the weight is "solely". The law targets decisions taken entirely by the machine, with no human judgement in the middle. And that is the practical key for an SME, because most AI uses in a small business are not of that kind. A chatbot that collects the enquiry and passes it to you decides nothing relevant: you prepare the reply. An automation that sorts invoices or classifies emails does not decide anyone's rights either. Article 22 does not complicate your life as long as the AI is an aid and the important decisions are still taken, or at least reviewed, by a person.

The problem appears if you hand over entirely to the system a decision that weighs on someone. The simple rule so you do not have to be a lawyer is this: the more important the decision is for the person, the more present the human eye has to be. For what really matters, let the AI propose and let someone of flesh and blood decide, or validate.

The EU AI Act: saying that people are talking to a machine

Besides the GDPR, which protects data, there is a new rule that regulates AI itself: the EU Artificial Intelligence Act (Reglamento Europeo de Inteligencia Artificial), known as the AI Act. According to the European Commission's regulatory framework for artificial intelligence, it entered into force on 1 August 2024 and applies in phases: prohibited practices from 2 February 2025, and the transparency obligations for chatbots and AI-generated content from 2 August 2026. The heavy high-risk obligations were postponed with the Digital Omnibus; the full timetable, with its updated dates, is in the guide to the EU AI Act for SMEs.

For a small business, the part that touches it most closely is precisely that transparency. From August 2026, if you have an AI system that converses with people (the website chatbot, the WhatsApp assistant, the agent that answers the phone), you have to make clear that whoever is on the other side is an artificial intelligence system and not an employee, unless it is obvious from the context. And if you generate content with AI, in certain cases that must also be flagged.

It is not a heavy burden. In a chatbot it is solved with a line at the start of the conversation saying it is an automated assistant. What the law aims at is that nobody believes they are talking to a person when they are not. Seen that way, telling people is not a commercial obstacle, it is basic honesty with the customer, and it also builds trust: people forgive being served by a machine; what they do not forgive is feeling deceived.

What can an SME do well with AI and the GDPR?

After so many rules, the reassuring part: complying is perfectly possible without being a large company with a legal department. It boils down to five habits.

  • Do not train on customer data without a clear basis. Using data to provide the service the customer asked for is one thing; using it to feed or tune a model is a different purpose that needs its own justification. If you do not have it, do not do it.
  • Put in writing which data does not go into an AI. A one-page company AI use policy stops someone pasting customer data into an unauthorised tool.
  • Say that AI is involved. Let the customer know when they are talking to an automated system and what is done with what they tell it. Reflect it in your privacy policy and, in conversational assistants, say it at the start.
  • Let a human review what matters. Let the AI prepare, sort and propose; let decisions that weigh on a person be taken or validated by someone. That keeps you on the right side of Article 22 almost effortlessly.
  • Process data in line with the GDPR, with or without AI. Collect only what is necessary, keep it for the right time, protect it and have a real way to honour the rights of access, rectification and erasure when someone asks.

None of these four points requires expensive technology or a law firm on the payroll. It requires designing the tool with them in mind from the start, which is much cheaper than fixing it afterwards. When we build an AI system for a business, these conditions are not a final add-on: they are part of how the project is framed.

Frequently asked questions

Can I use a chatbot with my customers' data without asking all of them for permission?

It depends on what for. If the chatbot processes the data within the service the customer has already asked you for (answering their enquiry, coordinating an appointment, following up a quote), you normally rely on performance of the contract or on legitimate interest, and you do not need separate consent for each. What you must always do is inform people that this processing exists. If you are going to use the data for a new purpose, such as training a model or different marketing, that changes things and the basis should be reviewed in detail.

Do I have to tell people they are talking to an AI?

Yes, and with ever more reason. The GDPR already pushes you to inform transparently, and the EU AI Act makes that notice an obligation from August 2026 for systems that converse with people. It is easy to comply with: it is enough for the assistant to say at the start that it is automated. It also works in your favour, because the customer relaxes when they know what they are talking to.

Can AI make decisions about my customers on its own?

For almost everything an SME does, AI does not make weighty decisions: it gathers information, sorts it and leaves it ready for you. Article 22 of the GDPR comes into play when a decision is taken solely by automated means and significantly affects the person. If you let a human review or validate those important decisions, you fall outside that case and avoid the problem at its root.

Do I need to change my privacy policy if I add AI?

Very probably, yes. Your privacy policy has to reflect what you really do with the data, and if you bring an AI system into customer service or your processes, that is new information the user has a right to know. It does not have to become a treatise; it has to be truthful and up to date.

If you are thinking of bringing AI into your business and you worry about doing it properly with your customers' data, tell us what you need and we will say whether we can do it, how we would approach it and what it would cost. If it is not for us, we will tell you that too. You can write to us from the contact page.

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.