AI in charities: the question is no longer whether we use it, but whether we use it well.
Artificial intelligence is already embedded in the charity sector.
Staff are using it to draft reports, summarise meetings, analyse information, prepare funding applications, develop communications and research potential supporters. Increasingly, AI is also appearing within the software charities already use.
Indeed, research cited by the Fundraising Regulator suggests that 76% of charities were using AI in 2025.
So, for trustees and charity leaders, the question is no longer simply:
“Should we allow people to use AI?”
A much better question is:
“How do we make sure AI is being used safely, appropriately and transparently?”
The risks are real
AI can be extraordinarily useful. I use it myself. But it should be treated as a tool, not as an authority.
There are several risks charities need to understand.
Accuracy. AI can produce information that sounds completely convincing but is wrong. A fabricated statistic or inaccurate claim in a funding application, trustee paper or supporter communication remains the charity's responsibility.
Confidentiality and data protection. Staff may inadvertently put personal, sensitive or confidential information into an AI system without understanding where that information goes or how it may subsequently be used. The ICO makes clear that organisations using AI still need to apply UK data protection principles.
Bias. AI systems learn from enormous quantities of existing information. That information inevitably contains biases. This becomes particularly important if AI is being used to help make decisions affecting beneficiaries, employees, volunteers or supporters.
Over-reliance. A polished answer is not necessarily a good answer. There is a danger that speed and convenience gradually replace judgement, professional expertise and proper scrutiny.
Authenticity. Charities depend heavily on trust. If a supporter believes they are reading the words of a beneficiary, fundraiser or CEO when they are actually looking at substantially AI-generated material, does that matter? In some circumstances, I think it does.
Accountability remains fundamental. You cannot delegate responsibility to an algorithm.
The Fundraising Regulator's guidance is particularly clear: charities remain accountable for fundraising activity involving AI, including AI-generated outputs and AI used by third-party fundraisers.
But banning AI isn't the answer
I don't think charities should respond by attempting to prohibit AI.
Apart from losing some potentially significant benefits, it may simply drive its use underground.
A member of staff who has discovered that AI can turn an hour's work into twenty minutes is unlikely to forget that capability because their organisation hasn't yet developed a policy.
The better approach is proportionate governance.
For most small and medium-sized charities, this doesn't need to mean creating another 40-page policy that nobody reads.
For a start here are seven practical steps:
1. Find out what is already happening.
Ask staff and volunteers what AI tools they are using and what they are using them for. Make this a learning exercise, not an investigation.
You may be surprised by the answer.
2. Agree a simple AI policy.
Set out what people can use AI for, what they shouldn't use it for and when additional approval is required.
The Fundraising Regulator now specifically recommends developing and agreeing an AI policy before using AI for fundraising.
3. Create some clear red lines.
For example, don't enter confidential, commercially sensitive, beneficiary or supporter information into publicly available AI tools unless you have specifically established that doing so is appropriate.
AI should not be making significant decisions about people without appropriate safeguards and human involvement.
And AI-generated facts, statistics, quotations and references should be checked before publication.
4. Train people rather than simply issuing a policy.
A one-hour practical session explaining what AI can do, what can go wrong and how to use it responsibly may achieve far more than circulating a policy by email.
Include volunteers and trustees where appropriate.
5. Keep humans accountable.
Someone should always remain responsible for the final output.
If I use AI to help draft a report, it is still my report. I am responsible for checking it, challenging it and deciding what eventually appears under my name.
The Fundraising Regulator similarly recommends human checks for accuracy, fairness and legality before AI-generated fundraising content is used.
6. Give trustees oversight, not operational control.
Trustees don't need to approve every use of Claude, ChatGPT, Microsoft Copilot or which ever AI the charity uses.
They do need assurance that the organisation understands where AI is being used, has considered the risks and has appropriate safeguards in place.
AI should therefore appear periodically on the board's risk and governance agenda.
7. Keep reviewing it.
AI is developing too quickly for a policy written today to sit untouched for three years.
Review your approach regularly, and whenever you introduce a significant new use of AI.
And then tell your supporters
This may be the area charities are most likely to overlook.
Trust is one of our sector's most valuable assets. Transparency about AI should therefore be seen as part of maintaining that trust rather than merely another compliance exercise.
The Fundraising Regulator recommends publishing an accessible policy explaining how AI is used in fundraising, particularly where AI-created material could be mistaken for real people or situations, where personal data is involved, or where AI chatbots are being used.
A charity's public statement doesn't need to be complicated. Something along these lines would be a reasonable starting point:
How we use artificial intelligence
We may use AI tools to help our staff work more efficiently, including supporting research, analysis, administration and the preparation of communications.
We remain responsible for everything we publish and for the decisions we make. Important AI-generated information is reviewed by a member of our team, and we do not rely on AI to replace professional judgement or appropriate human decision making.
We take particular care with personal, confidential and sensitive information and use AI in accordance with our data protection responsibilities.
Where the use of AI could materially affect how you understand a communication, image, interaction or decision, we will aim to make that clear.
That final sentence is important. I don't think every email needs to carry a declaration saying “AI was used in preparing this message.” That could quickly become meaningless.
Transparency should be proportionate. Using AI to help restructure a routine internal document is very different from creating an artificial image of a beneficiary, generating a supposedly personal supporter story, operating a chatbot that someone believes is a person, or using AI to influence a decision about an individual.
The greater the possibility of someone being misled or materially affected, the greater the need for transparency. That broadly reflects the approach now being taken by the Fundraising Regulator. This is ultimately about governance, not technology
Charity boards don't need to become experts in large language models. But trustees should be able to ask:
- Where are we using AI?
- What information are we putting into it?
- What decisions is it influencing?
- Who checks the output?
- Could anyone be harmed or misled?
- Would we be comfortable explaining our use of it publicly?
Those are governance questions, not technical ones.
AI offers charities enormous opportunities. Used intelligently, it can reduce administration, improve analysis and free people to spend more time on the human relationships at the heart of charitable work, but efficiency without accountability is a poor bargain. The aim should not be to become an AI-driven charity. It should be to remain a human led charity that knows how to use AI well.