Skip to content

Guide · Data protection

Can staff put client data into ChatGPT?

Using AI at work is allowed under UK GDPR. Putting client personal data into the wrong account isn't. What small firms need to know, in plain English.

By Stuart Kinsey, Co-founder of SimpleKPI4 min read

A prompt box asking "Can I put my client data into ChatGPT?" in front of the Saiela mark drawn in ribbons of violet, blue and cyan

It's Tuesday afternoon. Someone on your team pastes a client's email into ChatGPT to get a reply drafted. Is that a problem? Usually not. Sometimes, very much so.

The short answer: yes, your team can use ChatGPT and other AI tools at work. The risk comes from putting personal data into tools without the right protections, and the biggest problem is free and personal accounts.

This guide explains why, what counts as personal data, and what UK data protection law expects of a small firm. It's general information, not legal advice.

What counts as personal data

Personal data is anything about a living person who can be identified. That includes the obvious things, such as a name, email address, phone number, National Insurance number or bank details. It also includes a mix of details that together point to one person, such as a job title, a town and a date of birth.

In everyday work, that means a client's name in an email you want help drafting, a payslip in a spreadsheet you want to check, or a tenant's address in a letter.

Some data needs extra care. UK GDPR calls it special category data: health, ethnicity, religion, sexuality, political views, trade union membership, genetic and biometric data. Criminal records are treated in the same way. A note that a client claims tax relief because of a disability is special category data.

Why free and personal AI accounts are the problem

When someone uses a personal ChatGPT, Copilot, Claude or Gemini account for work, your firm has no contract with the AI provider. Many of those accounts can also use what's typed to train the provider's AI, unless someone has changed a setting.

Free or personal account Business plan
Contract with your firm None Yes, with a data processing agreement
Trains its AI on what's typed Often, unless switched off No, by default
Who decides what happens to the data The AI provider Your firm

Business plans such as ChatGPT Business, Microsoft Copilot signed in with a work account, Claude Team and Gemini in Google Workspace come with commercial terms, and the provider acts on your instructions. We compare each provider's plans in free vs business AI plans.

Not sure which accounts your team is actually using? AI used for work without the firm knowing is often called shadow AI, and our guide to finding shadow AI in a small firm shows you how to find it this week.

Is pasting client data into AI a data breach?

Under UK GDPR, sending personal data to someone who shouldn't have it is a personal data breach. In our reading, a free AI provider with no contract with your firm is exactly that. The Information Commission (ICO) hasn't ruled on this specific situation yet, so treat it with care rather than assume it's fine.

If it does happen:

  • Record it. Every personal data breach must be recorded, including the ones you don't need to report.
  • Report it if there's a risk to people. If the breach is likely to put people's rights and freedoms at risk, you must tell the ICO without undue delay, and where possible within 72 hours of finding out (ICO guide to personal data breaches). Our breach deadline calculator works out the deadline for you.
  • Tell the people affected if the risk is high.

What the law expects of a firm using AI

None of this means banning AI. It means using it the way you'd use any other supplier that handles client data:

  1. A lawful reason to use the data, usually legitimate interests, written down.
  2. A DPIA, a data protection impact assessment. The ICO says that in the vast majority of cases, using AI involves processing likely to result in a high risk.
  3. Contracts with providers: business plans with a data processing agreement.
  4. Records of processing that list your AI providers as recipients.
  5. Safeguards for transfers abroad, because most AI providers process data in the US.
  6. Telling people: your privacy notice mentions AI providers.
  7. Sharing only what's needed in each prompt.
  8. Training and rules: an AI policy your staff have read.

We go through each of these in what the ICO expects when your team uses AI.

Five things to do this month

  1. Choose business plans for the AI tools your team actually uses, and stop client work on free accounts.
  2. Write an AI policy, publish it and ask staff to acknowledge it.
  3. Put protection on every computer, after telling staff what it records.
  4. Complete a DPIA, and add your AI providers to your records of processing and privacy notice.
  5. Keep a breach register, including near misses.

Where Saiela fits

Saiela is a browser extension and console built for exactly this. It swaps personal details for placeholders before a prompt reaches the AI tool and puts them back in the answer, so your team can keep using AI on client work. It shows each AI tool's facts by plan so you can approve the right ones, and it drafts your AI policy on every plan, and on Team the DPIA and records of processing, from how your team really works. See how it works.

This guide is general information, not legal advice. For advice on your situation, speak to your data protection lead or a solicitor.

A laptop showing an AI chat where a file called All_Staff_Salaries.xlsx has just been sent, and the assistant replies "Well, this is awkward."

GuideData protection

Client data went into ChatGPT? What to do now

Someone put client data into ChatGPT? Don't panic: find out what was shared and on which account, then decide if the ICO needs telling. Step by step.

Stuart Kinsey3 min read

All posts

Want this handled for you?

Saiela does most of this automatically. Try it free for 7 days.