Guide · Data 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.
By Stuart Kinsey, Co-founder of SimpleKPI3 min read

Don't panic. First find out what was shared and on which account, then decide whether it needs reporting. If it does, you may have only 72 hours from when you found out, so start today.
This guide walks you through it, step by step. It's general information, not legal advice.
Work out your deadline now: our free breach deadline calculator shows when you'd need to tell the Information Commission (ICO), helps you decide whether it needs reporting, and lists what to record.
Step by step
- Find out what was shared, and when. Ask the person who did it, without blame. Which client? What details: names, addresses, bank details, National Insurance numbers, anything about health? When did it happen, and when did your firm find out?
- Check which account it was. Was it a business plan with a contract with your firm, such as ChatGPT Business, or a free or personal account? This matters most. Our guide to whether ChatGPT is safe for client data explains the difference between the plans.
- Delete the conversation, and check the account's settings. On a personal account, check whether "Improve the model for everyone" was switched on. Write down what you did and when.
- Assess the risk to the people involved. Think about what could happen to them because of it. A name alone is very different from bank details or health information.
- Record it in your breach register. Every personal data breach must be recorded, including the ones you don't report (UK GDPR, Article 33(5)). Write down the facts, the effects and what you did about it.
- Report it to the ICO if it's likely to put people at risk. Do it without undue delay and, where possible, within 72 hours of becoming aware of it (Article 33(1)). If you report late, say why.
- Tell the people affected if the risk is high (Article 34). Explain what happened, the likely effects and what you're doing about it.
- Fix the cause. Usually that means a clearer AI policy, a short refresher for the team, and the right tools, so it doesn't happen again.
Is it always a breach?
Not necessarily, and it depends on the account. Under UK GDPR, sending personal data to someone who shouldn't have it is a personal data breach. In our reading, a free or personal AI account with no contract with your firm is exactly that. The ICO hasn't ruled on this specific situation yet, so treat it with care rather than assume it's fine.
If it went into a business plan your firm has approved, with a data processing agreement in place, the provider is acting on your instructions, and it's much less likely to be a breach. You should still record what happened and check it was within your policy.
We explain the reasoning in more detail in can staff put client data into ChatGPT?
How to stop it happening again
- Approve the right plans. Pay for business plans of the tools people actually use, and keep client work off personal accounts. Free vs business AI plans compares them.
- Write it down. Our free AI policy template covers what must never go into AI, personal accounts, and how to report a mistake quickly.
- Catch it before it's sent. Saiela checks what people type into AI tools on their own computer, and swaps client details for placeholders before anything is sent. It warns people on personal accounts, or blocks them, and the breach register is free on every plan.
Sources
- Information Commission (ICO): Personal data breaches: a guide
- UK GDPR: Article 33, notifying the ICO and Article 34, telling the people affected
- OpenAI: How your data is used to improve model performance
Questions
The short answers.
Do I have to report it to the ICO?
Only if the breach is likely to result in a risk to people's rights and freedoms. If it is, report it without undue delay and, where possible, within 72 hours of becoming aware of it. Either way, you must record it in your breach register.
Can I get the data deleted from ChatGPT?
Delete the conversation, and keep a note that you did. Whether and when that removes it from OpenAI's systems depends on OpenAI's terms for that plan, so check its help pages for how long deleted chats are kept. On a personal plan with "Improve the model for everyone" switched on, the conversation may already have been used for training.
Should I tell the client?
If the breach is likely to result in a high risk to the people involved, UK GDPR says you must tell them without undue delay. Even when it isn't required, many firms choose to, because it's the honest thing to do and protects the relationship. Record what you decided and why.
General information, not legal advice.

