Black Friday10% off any legal service

Can I Use Customer Data With AI?

byAlex Solo9 min read

Priya runs a small online homewares shop in Leeds. Her team wants to use an AI tool to draft product recommendations for customers based on their past purchases, and someone suggests pasting a spreadsheet of customer names, emails and order history into the tool to get ideas.

It sounds efficient. It might even be genuinely useful. But Priya pauses, because that spreadsheet is customer data, and once it goes into an AI tool it can be hard to know exactly where it ends up.

The good news is that using AI with customer data isn't automatically off limits in the UK. You just need clear rules, the right privacy checks, and a practical AI governance policy for your team before anyone starts pasting customer details into a chatbot.

Can I Use Customer Data With AI Tools?

Yes, you can use customer data with AI tools if you have a lawful basis under UK GDPR, you use the data fairly and transparently, and the tool is set up with proper data protection and information security controls.

For UK businesses, the key laws are the UK GDPR and the Data Protection Act 2018. If you're using AI for email marketing or electronic communications, the Privacy and Electronic Communications Regulations, often called PECR, may also apply.

The first question is whether the information is personal data. If the data identifies a living person, or could identify them when combined with other information, UK GDPR is likely to apply. Names, email addresses, phone numbers, customer IDs, order histories, support tickets and loyalty profiles can all be personal data.

You also need to know your role:

  • You're usually the controller if you decide why and how customer data is used with the AI tool.
  • The AI provider may be a processor if it only handles the data on your instructions.
  • The provider may be a separate controller if it uses your data for its own purposes, such as improving or training its general model.

That role matters because it affects what contracts, privacy notices and customer information you need. If a tool's terms say it can use prompts and uploads to train its services, that may be a poor fit for confidential customer data unless you have a clear lawful basis and have told customers properly.

What Counts As Customer Data In An AI Tool?

Customer data is any information your business holds about customers or potential customers, and it can become personal data if it relates to an identifiable person.

This includes obvious details like:

  • Names, addresses, phone numbers and email addresses
  • Account logins, customer numbers and loyalty IDs
  • Purchase history, refunds and delivery details
  • Customer service messages, complaints and call notes
  • Survey responses, reviews and preference data
  • Website behaviour linked to a customer profile

Some data needs extra care. Under UK GDPR, special category data includes information about health, ethnicity, religious beliefs, political opinions, sex life or sexual orientation, biometric data used for identification, and some trade union information. If your AI use involves this kind of data, you'll usually need a stronger reason, tighter controls and often a Data Protection Impact Assessment.

There's also a practical difference between anonymised and pseudonymised data. Anonymised data can't reasonably identify someone, so UK GDPR doesn't apply to it. Pseudonymised data, such as replacing names with customer IDs, can still be personal data if you or someone else can re-identify the person.

Say Priya wants AI to help plan a customer email campaign. Uploading a spreadsheet with 2,000 customer names, emails, order dates and average spend creates a privacy risk. A safer option may be to ask the AI tool to work from non-identifying summaries, such as 350 repeat candle buyers, average order value £42, most popular scent vanilla, no names or emails included.

Can My Team Put Customer Data Into ChatGPT Or Similar Tools?

Your team can only put customer data into ChatGPT or similar AI tools if you've approved the tool, checked its terms and privacy settings, and given staff clear instructions about what they can and can't enter.

This is where many businesses get caught out. A staff member might paste a customer complaint, a sales lead list or a support chat into an AI tool to draft a reply. They may be trying to help, but they could accidentally disclose personal data or confidential business information to a third party.

Before staff use any AI tool with customer information, check:

  • Whether the tool stores prompts, files or outputs
  • Whether your inputs can be used to train or improve the model
  • Where the data is hosted or accessed from
  • Whether you can delete inputs and outputs
  • Whether the provider offers business settings that turn off training on your data
  • Whether there's a data processing agreement that meets UK GDPR Article 28 if the provider is your processor

For everyday use, many businesses choose a simple rule: staff must not enter identifiable customer data into public AI tools unless the tool has been approved for that purpose.

You can still get value from AI without exposing customer details. Staff can ask for help drafting a response using a fictional version of the facts, or remove names, email addresses, order numbers and other identifiers before using the tool. If the matter is sensitive, confidential or unusual, it should stay out of the tool unless you've done a proper review.

What Privacy Checks Should I Do Before Using AI With Customer Data?

You should do a privacy review before using AI with customer data, especially if the tool will process personal data, make recommendations about customers, or support marketing decisions.

A practical review doesn't have to be huge. For a small business, it can be a clear record of what you're using, why you're using it, what data goes in, what comes out, and how you're controlling the risks.

AI Privacy Review Checklist

  • Purpose: Write down what the AI tool will do, such as drafting replies, segmenting customers or summarising feedback.
  • Data: List the exact categories of customer data involved.
  • Lawful basis: Identify your UK GDPR lawful basis, such as contract, legitimate interests or consent, depending on the use.
  • Customer notice: Check whether your privacy policy or notice tells customers about this type of processing.
  • Provider terms: Review whether the AI provider uses your data for training, product improvement or its own purposes.
  • Security: Confirm access controls, encryption, user permissions, information security rules and deletion options.
  • International transfers: If data leaves the UK, check the transfer mechanism, such as adequacy regulations, the UK International Data Transfer Agreement or the UK Addendum to EU standard contractual clauses.
  • Retention: Decide how long prompts, files and outputs are kept.
  • Human review: Make sure a person checks AI outputs before they're used with customers.
  • DPIA: Consider whether you need a Data Protection Impact Assessment, particularly for high-risk processing, profiling or special category data.

If you're using AI for marketing, don't forget PECR. For example, email and SMS marketing often needs consent unless a narrow soft opt-in applies. AI doesn't remove those rules. It just adds another layer of data use to check.

You should also keep records. If the Information Commissioner's Office ever asks how you assessed a tool, a short, dated privacy review is much better than trying to piece it together later from memory.

How Do I Write An AI Acceptable Use Policy?

Your AI acceptable use policy should tell your team which AI tools are approved, what information they can enter, what is banned, and who to ask when they're unsure.

This is different from a privacy policy. Your privacy policy tells customers what you do with their personal data. An AI acceptable use policy is an internal document for staff, contractors and managers. It sets everyday rules so people don't have to guess.

Here is a practical outline you can adapt, but have a lawyer tailor it before you use it. A template may miss the way your business actually handles customer data, confidential information, marketing and staff access.

  1. Purpose: Explain why your business uses AI and that staff must follow the policy when using AI for work.
  2. Approved Tools: List the AI tools staff can use and whether they're approved for personal data, confidential data or general drafting only.
  3. Prohibited Use: Ban staff from entering customer data, passwords, payment details, special category data or trade secrets into unapproved tools.
  4. Permitted Use: Give examples of safe uses, such as brainstorming non-confidential content, summarising anonymised feedback or drafting first versions for human review.
  5. Customer Data Rules: Set out when customer data can be used, who can approve it, and what must be removed first.
  6. Confidentiality: Remind staff that business secrets, client materials and supplier information may be confidential even if they aren't personal data.
  7. Human Review: Require staff to check AI outputs for accuracy, tone, bias, privacy and legal risk before using them.
  8. Marketing Use: Set rules for AI-assisted emails, targeting, customer segmentation and claims about products or services.
  9. Record Keeping: Tell staff when to save prompts, outputs, approvals and privacy checks.
  10. Incidents: Explain what staff must do if they accidentally enter customer data into the wrong tool.
  11. Training And Updates: Say how staff will be trained and how often the policy will be reviewed.

Your employee use of AI policy should sit neatly with your employment contracts, staff handbook, information security policy, confidentiality clauses and data protection procedures. If those documents say different things, staff may follow the wrong rule.

What If Customer Data Has Already Been Put Into An AI Tool?

If customer data has already been put into an AI tool, act quickly, contain the issue, and work out whether it's a personal data breach under UK GDPR.

Start with the facts. Ask what data was entered, which tool was used, who had access, whether the data was stored, whether it can be deleted, and whether the provider can use it for training. Keep a clear written record of what happened and what you did next.

A sensible first response usually looks like this:

  1. Stop further use: Tell staff not to enter any more customer data into that tool until you've reviewed it.
  2. Preserve evidence: Save the prompt, uploaded file name, time, user and any output if you can do so safely.
  3. Check deletion options: Remove the data from the tool where possible and ask the provider to confirm deletion or isolation.
  4. Assess the risk: Consider the sensitivity of the data, number of people affected, chance of misuse and whether the provider can access or reuse it.
  5. Decide on notifications: If it's a personal data breach that's likely to risk people's rights and freedoms, you generally need to notify the ICO within 72 hours of becoming aware of it.
  6. Update your controls: Fix the gap with training, tool settings, access controls or an AI acceptable use policy.

Not every mistake needs to be reported to the ICO or customers. For example, if one staff member pasted a customer's first name and a non-sensitive draft email into an approved business AI tool with training switched off, the risk may be low. If someone uploaded a full customer database with names, emails, addresses and complaint notes into an unapproved public tool, you should treat it much more seriously.

You may also need to think about customer rights. Customers can make subject access requests, and the usual UK GDPR response period is one month. If AI tools store prompts or outputs containing personal data, you need a way to find and manage that information.

Key Takeaways

  • You can use customer data with AI tools, but only if you comply with UK GDPR, the Data Protection Act 2018 and any marketing rules that apply.
  • Customer data includes more than names and emails. Order history, support notes, IDs and behavioural data can all be personal data.
  • Staff shouldn't put customer data into ChatGPT or similar tools unless the tool has been approved and you've checked the settings, terms and data processing position.
  • An AI acceptable use policy gives your team practical rules on approved tools, banned data, human review and what to do if something goes wrong.
  • A privacy review, data processing agreement and clear customer notice can help you use AI confidently without creating avoidable risk.

If you'd like help with your AI acceptable use policy, privacy review, data processing agreements and customer data controls, you can reach us at 08081347754 or team@sprintlaw.co.uk for a free, no-obligations chat.

Get your customer-facing terms right

What should your privacy and online terms cover?

If you collect customer data, sell online or run marketing campaigns, your public terms and privacy documents should match the real customer journey.

Alex Solo
Alex SoloCo-Founder

Alex is Sprintlaw’s co-founder and principal lawyer. Alex previously worked at a top-tier firm as a lawyer specialising in technology and media contracts, and founded a digital agency which he sold in 2015.

Get your customer-facing terms right

Get in touch with our team

Tell us what you need and we'll come back with a fixed-fee quote - no obligation, no surprises.

Need support?

Need help with your business legals?

Speak with Sprintlaw to get practical legal support and fixed-fee options tailored to your business.