AI for charities: five uses that don't need a big budget

You don’t need a big budget to get real value from AI as a charity. The best places to start are everyday jobs: first drafts, summaries, answering the same questions again and again, admin, and making information easier to read. None of them needs custom software to get going, but all of them need a person checking the output. You’d be in good company: almost eight in ten charities (79%) in the 2026 Charity Digital Skills Report said they were using AI tools in some way, and more than half (56%) named skills and technical expertise as the biggest barrier.

1. First drafts of funding applications

AI is good at turning your notes, past bids and impact reports into a first draft that fits a word limit, and at checking a draft against a funder’s criteria. It’s bad at knowing your community, and it leans towards the generic phrases every assessor has read a hundred times.

Funders are starting to publish their own rules, and they don’t all say the same thing. The National Lottery Community Fund says you can use AI tools to help write an application and that it won’t reject one just because AI was used. It also warns that AI-supported applications “do not tell the unique story of your community”, that free tools may store what you type, and that AI-suggested budgets need checking. Historic Environment Scotland expects every application to have a human author and cautions against using AI heavily. So check each funder’s guidance before you start, and if a funder says nothing, ask.

2. Summarising consultation and survey responses

AI can sort hundreds of free-text answers into themes, count how often each comes up and pull out quotes that show each one. Treat that as a first pass, not the analysis. When the UK government’s Consult tool was trialled on more than 2,000 responses to a Scottish Government consultation on non-surgical cosmetic procedures, experts in the Scottish Government checked and refined the themes it found, and officials read every response as well, to test it. A lighter version works for a charity: remove names and identifying details first, read a sample of the raw responses against the summary, and look for the minority views a summary can smooth away.

3. Answering common questions from your own documents

“How do I set up a regular gift?” “Can I volunteer at 16?” “Is the shop open on Sundays?” An assistant that searches your own documents and answers from them, with a link to the source, can handle routine questions like these and pass the rest to a person. We explain the approach in how retrieval works. Start with public information such as FAQs, published policies and event pages. If people might contact you in distress, the assistant must hand them to a person straight away, and you should test that before anything else.

4. Volunteer rotas and admin

Rotas, reminders and meeting notes can eat a surprising amount of a small team’s week. AI can turn a jumble of availability replies into a draft rota, write the reminder messages and turn meeting notes into a list of actions. Check the rota against the original replies, because a language model can drop a name or double-book someone without noticing. Volunteers’ contact details are personal data, so use a business account whose terms you’ve checked. If you want to automate one of these jobs properly, start with something boring.

5. Accessible versions and translations

AI can draft plain-English versions of dense documents, short summaries, starting points for Easy Read versions (simple words with supporting pictures, often used by people with learning disabilities) and translations into the languages your community uses. Every one needs a human check. A fluent speaker, ideally from the community, should check translations. People who use Easy Read should test those versions. And don’t rely on an unchecked translation for anything about health, safety, money or legal rights.

Beneficiary data needs extra care

Charities often hold information about people at difficult moments in their lives. Some of it will be special category data under UK GDPR, which includes health, racial or ethnic origin, religious or philosophical beliefs, sex life and sexual orientation. To use that kind of data at all, you need both a lawful basis under Article 6 and a separate condition under Article 9. The ICO says that “in the vast majority of cases” using AI will involve processing likely to result in a high risk, which means doing a data protection impact assessment (DPIA) first. It also says a DPIA is required if you plan to use special category data to decide who gets access to a service or benefit.

Our UK GDPR checklist for AI tools walks through the steps.

Help in Scotland

• SCVO (the Scottish Council for Voluntary Organisations) runs digital training and events for the voluntary sector, including its DigiShift webinars, and offers a Digital Check-up tool.
• Business Gateway offers free one-to-one advice, webinars and events, including introductions to AI. It’s aimed at businesses, so ask what it can offer a charity.
• The Data Lab’s National AI Adoption Programme is a programme for SMEs (small and medium-sized enterprises). Its page doesn’t say whether charities qualify, so ask before you count on it.

There’s more on paying for AI work in funding AI projects in Scotland.

Start with a policy

Before any of this, agree a short AI policy: which tools are allowed, what must never go into them, and who checks what. Our guide to an AI policy your team will read has more. If one of these five fits your charity, tell us which.