Is it safe to put donor data into ChatGPT?

No. Donor records, giving history and contact details are personal data under UK GDPR, and much of it is also governed by the Code of Fundraising Practice. It should never go into the free or consumer version of ChatGPT, which may use inputs to train its models. Use it only in tools formally approved for personal data, under a data processing agreement.

What can charities safely use ChatGPT for?

Plenty: drafting non-sensitive communications, summarising public documents, brainstorming campaign ideas, writing job adverts, and turning rough notes into readable copy, as long as nothing personal or confidential goes in. The rule of thumb is simple: if it names or identifies a real donor, beneficiary or staff member, keep it out.

Resources · 13 July 2026

Can charities use ChatGPT? Donor and beneficiary data rules (UK)

Q: Can charities use ChatGPT?

Yes. There is no rule stopping a UK charity from using ChatGPT or similar tools, and many already do for drafting, research and admin. The care is all about the data: never put donor, beneficiary or other personal and confidential information into a consumer AI tool, and have a short written policy so staff and volunteers know where the line is.

By the Operating Bench Team · Last reviewed 13 July 2026

Short answer: yes, charities can use ChatGPT, and a lot of good work is already being done with it, from drafting grant bids to summarising reports. The question that actually matters is not whether you may use it, but what you must never put into it. Here is the practical line for a UK charity.

What charities can safely use ChatGPT for

Used with public or non-sensitive information, ChatGPT is a genuine time-saver for a stretched team:

Drafting and improving communications, from newsletters to social posts, where no personal data is involved.
Summarising public documents, reports and guidance.
Brainstorming campaign ideas, event plans and fundraising angles.
Writing first drafts of job adverts, role descriptions and policies.
Turning rough notes into clear, readable copy.

In all of these, the input is either public or generic. That is the safe zone.

What must never go in

The risk with a consumer AI tool is simple: what you type in can leave your control, and the free version may use your inputs to train the model. So the hard line for a charity is:

Donor, supporter and beneficiary data, and anything confidential, must never go into the free or consumer version of ChatGPT.

That means no names, no giving or Gift Aid history, no contact details, no case notes, and absolutely nothing about a vulnerable beneficiary. Donor records and giving history are personal data under UK GDPR, and they also sit under the Code of Fundraising Practice. Beneficiary and safeguarding information is among the most sensitive data any organisation holds, and the least suited to a consumer tool.

Donor research and wealth screening deserve a special mention. They are tempting to hand to an AI, but profiling donors engages both UK GDPR and the fundraising rules: be transparent about how you use data, respect donors’ preferences, and never let a tool push you into undue pressure.

How to use it safely: the approved-tool approach

The way through is not to ban AI, which just drives it underground onto personal accounts. It is to give people a clear, approved way to use it:

Approve specific tools for specific data. The free consumer ChatGPT is fine for public and non-personal work. For anything touching personal data, use a version with proper terms, a business or enterprise plan under a data processing agreement that does not train on your inputs, and only where you have assessed it.
Bring volunteers into scope. Much charity work runs on volunteers using personal devices and accounts, so any rule that only reaches employees misses half the risk. The same short rules should apply to everyone acting for the charity.
Keep a human in the loop. AI is confidently wrong sometimes. A person should check anything before it goes to a donor, funder or beneficiary, and that person is accountable, not the tool.
Write it down. A one-page policy that names what is allowed, what is not, and who to ask turns all of this into something staff and volunteers can actually follow. Trustees increasingly expect to see it, too.

Where this fits with your wider duties

Using ChatGPT well is one piece of a charity’s AI governance. The fuller picture, covering donor data, fundraising, safeguarding and trustee oversight, is in our guide to an AI policy for charities. The principle is the same throughout: keep the time savings, protect the people whose data you hold.

Getting a policy in place

You can set your own rules from the guidance above. To move faster, the free AI Starter Kit gives you a one-page set of ground rules and a quick guide to what data is safe to put into AI tools. For the complete system, the AI Safe-Use Pack for charities is written specifically around donor and beneficiary data, the Code of Fundraising Practice and trustee oversight, with an editable policy, an approved-tools matrix and a risk register built for how charities actually work.

Whichever way you go, the message to your team is the good-news version: yes, use AI, here is how to do it without putting anyone’s data or trust at risk.

This article is general information, not legal advice. For advice on your specific circumstances, consult a suitably qualified professional.