AI in charities and fundraising: trustees, donors and honest appeals

Cue 1 Two regulators' documents side by side, dated 2 April 2024 and 8 December 2025

EMMA 00:00 George, I'm a trustee of a small charity. Our fundraiser wants to use AI for the Christmas appeal. Is there a rule?

GEORGE 00:08 There's guidance, from the Fundraising Regulator, published on 8 December 2025. It doesn't add new rules. It shows how the Code of Fundraising Practice applies, so AI is used in a legal, open, honest and respectful way.

EMMA 00:23 And the Charity Commission?

GEORGE 00:25 It wrote about AI in April 2024. It said it didn't anticipate producing specific new guidance. But it was clear that trustees remain responsible for decision making, and that mustn't be delegated to AI.

Cue 2 A board table where one trustee holds all the knowledge, redrawn with it shared round the table

EMMA 00:38 So who's accountable if the appeal goes wrong?

GEORGE 00:42 The charity, and ultimately you, the trustees. The guidance says you're responsible and accountable for your use of AI for fundraising, including that of your third-party fundraisers.

EMMA 00:52 Only one of us on the board understands this stuff.

GEORGE 00:56 That's named in the guidance. Take all reasonable steps to avoid a situation where trustee understanding and knowledge about AI is concentrated among a minority.

Cue 3 A policy document placed before a tool on a timeline

EMMA 01:06 So what comes first?

GEORGE 01:08 A policy. Before the tool. Saying when AI creates or informs content, makes decisions or carries out tasks. And reviewed regularly.

EMMA 01:16 Do we publish it?

GEORGE 01:18 The regulator recommends you publish how you use AI in your fundraising, accessibly, on your website.

Cue 4 A charity website with a plain page headed how we use AI in our fundraising

EMMA 01:24 And then we can start?

GEORGE 01:26 After a risk assessment. The guidance says avoid using any AI tool before conducting one.

EMMA 01:32 We're four staff. How big does that need to be?

GEORGE 01:36 Proportionate. Tidying the wording of a newsletter needs less than a tool deciding whom to ask for money.

Cue 5 A one page risk assessment with five questions, scaled small for a newsletter and large for donor selection

EMMA 01:44 What do I ask?

GEORGE 01:45 What will the tool do? What goes into it, and where does that go? What could it get wrong, and who'd be affected? Who checks the result? And what do we tell donors?

EMMA 01:58 And once it's written?

GEORGE 02:00 Keep it with the policy. Look at both again whenever a tool, a supplier or a use changes.

Cue 1 A scale running from low to high risk of misleading, with tidied grammar at one end and a generated child at the other

EMMA 02:07 George, our appeal needs a photo and we don't have one. The tool can generate a child at a food bank. Can we use it?

GEORGE 02:17 Think about what a donor would believe. The guidance gives a sliding rule. The amount of transparency should be in proportion to the risk of misleading donors. The greater the risk, the more transparent you should be.

EMMA 02:32 And a generated child?

GEORGE 02:34 High risk. A donor will take it as someone you helped. If the person, the picture or the words were generated, they mustn't be left believing otherwise.

Cue 2 A person reading an appeal against three words: accuracy, fairness, legality

EMMA 02:44 What about AI tidying the grammar of a true appeal?

GEORGE 02:48 No label needed. Nobody's misled about anything.

EMMA 02:51 Who checks the appeal before it goes?

GEORGE 02:54 A person. The guidance says have a process for a human to check the accuracy, fairness and legality of any content generated by AI before using it.

Cue 3 A supporter list at a gate with three questions: use, store, share

EMMA 03:05 Break those down.

GEORGE 03:06 Accuracy. Are the figures and the claims about impact true? Fairness. Is it respectful, without undue pressure? Legality. Does it meet the Code, data protection law and the advertising rules?

EMMA 03:18 We'd like the tool to pick which donors to ask for more.

GEORGE 03:23 Then you're using personal data. The guidance says at all times know how any AI tool will use, store and share the material you put into it.

Cue 4 A tool choosing whom to ask, with an elderly donor protected by the same rule as for a human fundraiser

EMMA 03:34 Use, store, share.

GEORGE 03:35 Three verbs. If you can't answer them, the supporter list doesn't go in. And your privacy information should say how AI is involved.

EMMA 03:44 Some of our donors are elderly.

GEORGE 03:46 The Code's section on people in vulnerable circumstances applies to a tool that chooses whom to ask, how often and for how much. Exactly as it would to a person making those choices.

Cue 5 A chatbot and an agency both wearing the charity's name badge, with a monitoring log beside them

EMMA 04:00 We also have a chatbot on the donation page.

GEORGE 04:03 Where there's no immediate human oversight, the guidance says have measures to monitor and audit it. It speaks for the charity.

EMMA 04:12 And our telephone fundraising agency?

GEORGE 04:14 Same accountability. Review the agreement, ask how they use AI on your behalf, and monitor it. You answer for what's done in your name.