All articlesArtificial Intelligence

UK GDPR and AI Chatbots: A Practical Checklist for UK Businesses

CT
CreativeIn Tek Team
· 7 min read
UK GDPR and AI Chatbots: A Practical Checklist for UK Businesses

UK GDPR and AI chatbots are now a routine pairing for UK businesses: a chatbot on a website or WhatsApp number collects names, phone numbers, questions and sometimes sensitive details, and the business that runs it is responsible for how that personal data is handled. The Information Commissioner's Office (ICO) publishes guidance on AI and data protection, read in a browser on 17 November 2026, that sets out the questions a business should work through: who is controller and who is processor, whether a data protection impact assessment (DPIA) is needed, how people will be told, what lawful basis applies, how accurate and fair the system is, how it is secured and how individual rights will be met. This guide turns that structure into a practical checklist for a UK small business. The ICO says the guidance is under review because of the Data (Use and Access) Act, so check its current pages before launch. It is general information, not legal advice.

Key takeaways

  • The ICO's Guidance on AI and data protection was last updated on 15 March 2023, and the ICO says it is under review because of changes made by the Data (Use and Access) Act.
  • The ICO says that in the vast majority of cases the use of AI involves processing likely to result in a high risk to individuals' rights and freedoms, which triggers the legal requirement to carry out a DPIA.
  • If a business decides a particular AI use is not high risk, the ICO says it must still document how it made that assessment.
  • If a DPIA shows residual high risk that cannot be reduced, the ICO says the business must consult the ICO before starting the processing.
  • Businesses should work out whether they are controller, joint controller or processor for each organisation involved, because that decides which obligations apply to whom.

UK GDPR and AI chatbots: what the ICO guidance covers

The ICO's guidance is organised in chapters that make a ready-made agenda: accountability and governance, transparency, lawfulness, accuracy, fairness, security and data minimisation, and individual rights, plus an annex on fairness through the AI lifecycle and a risk toolkit. The landing page says the guidance was updated after requests from UK industry to clarify requirements on fairness and that it helps organisations adopt new technologies while protecting people. It carries a banner that the guidance is under review because of the Data (Use and Access) Act, and points to a page of plans for new and updated guidance. A business using a chatbot should read the chapters relevant to its use, and should expect some details to change.

Step 1: work out who is controller and who is processor

The ICO's accountability chapter says several organisations are often involved in developing and deploying AI systems that process personal data, that the UK GDPR recognises that they will not all have the same degree of control or responsibility, and that it is important to identify who is a controller, a joint controller or a processor so you understand which obligations apply to which organisation. For a small business, the usual picture is that the business decides why and how customer messages are used, which makes it a controller, while the chatbot platform and the underlying model provider may be processors or, in some cases, controllers for parts of their own processing. The contract and the provider's terms should say which. Ask the vendor to explain in writing how customer messages are used, whether they are used to train models and where they are stored.

Step 2: decide whether a DPIA is needed

The ICO says that in the vast majority of cases the use of AI will involve processing likely to result in a high risk to individuals' rights and freedoms, and will therefore trigger the legal requirement to undertake a DPIA, with the assessment made case by case. It adds that where a business concludes a particular use of AI is not high risk, it must still document how it reached that view, and that if the DPIA shows residual high risk that cannot be sufficiently reduced, it must consult the ICO before starting the processing. A simple FAQ chatbot that holds no personal data may be low risk, while a chatbot that handles health or financial details is more likely to need a full assessment. The practical step is to write the assessment down before launch.

Step 3: tell people, and choose a lawful basis

  • Transparency: the ICO has a chapter on transparency in AI. At a minimum, tell people at the start of the chat that they are talking to an automated system, what is collected and why, and link to the privacy notice.
  • Lawful basis: choose and record the basis for each purpose, such as answering an enquiry or arranging a booking. If the chatbot handles special category data such as health information, an additional condition is needed.
  • Marketing: do not turn chat data into marketing without a valid basis, and see the guide to WhatsApp for UK businesses for the electronic marketing rules.

Step 4: accuracy, fairness and human oversight

The ICO treats accuracy and fairness as separate chapters. For a chatbot that answers customer questions, accuracy means the information it gives is correct and kept up to date, so keep the knowledge base reviewed and test the chatbot on real questions. Fairness means it should not treat people worse because of who they are, so test it with varied phrasing and read conversations for bias. Where a chatbot's output affects a person significantly, for example refusing a service automatically, the ICO guidance includes a chapter on Article 22 of the UK GDPR on solely automated decisions, so keep a person in the loop for decisions that matter.

Step 5: security, data minimisation and individual rights

  • Collect less: do not ask for details the chatbot does not need, and avoid asking for sensitive information in a chat window.
  • Limit retention: decide how long conversations are kept and delete them on schedule.
  • Control access: restrict who can read transcripts and keep an audit trail.
  • Prepare for requests: be able to find, export and delete a person's chat data when they exercise their rights.
  • Check where data goes: know which countries the platform and the model provider process data in.

A launch checklist

  • Map the data: what the chatbot collects, where it goes and who can see it.
  • Identify roles: controller, joint controller or processor for each party, and put terms in writing.
  • Complete and file the DPIA, or document why none is needed.
  • Update the privacy notice and add a short notice at the start of the chat.
  • Test for accuracy and fairness on real questions before going live.
  • Set a human handover for complaints and important decisions.
  • Review quarterly, and re-read the ICO's pages for updates.

Where CreativeIn Tek fits

CreativeIn Tek builds AI agents through its AI solutions service and the AI Agent product, which answers customers on WhatsApp, linked by QR code. CreativeIn Tek does not claim that its products make a business compliant with UK GDPR, and a UK business must assess its own use. The difference between a chatbot and an agent is explained in the guide to AI agents versus chatbots, and a business can talk to CreativeIn Tek about its plans.

Frequently asked questions

Do I need a DPIA for an AI chatbot?

The ICO says that in the vast majority of cases the use of AI triggers the legal requirement for a DPIA, assessed case by case, and that if you decide it is not needed you must document how you decided.

Who is responsible for a chatbot's data protection?

The business that decides why and how personal data is used is usually the controller, but the ICO says to identify the role of each organisation involved.

What if the DPIA shows a high risk I cannot reduce?

The ICO says you must consult it before starting the processing.

Is the ICO's AI guidance current?

It was last updated on 15 March 2023 and the ICO says it is under review because of the Data (Use and Access) Act, so check the latest pages.

Can a chatbot make decisions about people?

The ICO has a chapter on Article 22 and solely automated decisions, so keep a person involved in decisions that significantly affect people.

Last reviewed: 17 November 2026, against the ICO's Guidance on AI and data protection and its chapter on accountability and governance in AI. The ICO says the guidance is under review. Laws and guidance change, and this article is general information and not legal advice.

ai

Enjoyed this article?

Subscribe for occasional deep dives on shipping software well.