Nonprofits are usually the teams with the most work and the fewest people to do it. That's exactly why AI is worth learning, and exactly why it needs care. A charity's reputation is built on trust, and one confident, wrong paragraph in an appeal or a report can undo far more than the hours it saved.

I've spent 23 years in marketing, I've worked with Greenpeace among many other organisations, and since 2024 I've rebuilt my work around practical AI. Here's how I'd approach it for a small nonprofit team.

Where AI helps a small nonprofit team

The best starting points are the jobs that repeat every week and drain staff and volunteer time.

• Supporter communications: first drafts of newsletters, thank-you emails and social posts, edited by someone who knows your supporters.
• Admin: meeting notes, volunteer rotas, policy summaries and replies to common questions.
• Reporting: turning your own notes and figures into a clear structure for trustees or partners.
• Bid and application writing: organising your evidence and drafting sections, with the specifics written by people who know the work.
• Research: a quick first pass on a topic, checked against the sources before anyone relies on it.

What the Charity Commission says trustees must keep

For charities in England and Wales, the Charity Commission has been clear. In its blog on charities and artificial intelligence, it says trustees remain responsible for decision making, so it's vital that process isn't delegated to AI or based on AI generated content alone. It adds that trustees may not be meeting their duties if a charity relied solely on AI advice for a critical decision without reasonable independent checks.

The same post suggests charities consider whether an internal AI policy would help, so it's clear how and when AI can be used in governance, by staff, and in services. Whatever country you work in, that's sound practice for any nonprofit board.

The generic-content trap

Here's the warning I'd give every nonprofit writer. In a later post, the Charity Commission says AI-generated content in charity registration applications is often too generic and fails to reflect the specific activities or aims of the organisation, leading to more rejections.

That's the same failure I see in appeals and bids written with AI. Readers and supporters can tell when a page could belong to any charity. Use AI to organise and draft, then put back the names, places, stories and specifics only your team knows.

A safe way to start

You don't need a big programme. You need a small, careful start that the whole team understands.

1. Agree a one-page policy: which tools, for which tasks, and what never goes in, especially beneficiary and donor details.
2. Pick two weekly tasks that eat staff time and build AI into those first.
3. Make a human check part of every workflow before anything is published or sent.
4. Train staff and volunteers on the tools they'll use, not on AI in general.
5. Review after a month and keep only what saved time without risk.

How I work with nonprofits

I run hands-on training and implementation for small teams, online or in person, in English. We start from your real tasks and build the workflows together, with the rules written in. There's a deeper guide to AI for charity grant writing and admin, a set of ChatGPT prompts for charity appeals, and the ways to work together are on my work with me page.

What AI shouldn't do for a nonprofit

It shouldn't make decisions about people, write your safeguarding policy, or speak for your beneficiaries. It shouldn't be fed personal details about the people you serve unless you're certain the tool and its settings allow it. And it shouldn't replace the human voice your supporters trust. Those lines are worth writing down before the first prompt.

Getting staff and volunteers on board

In most nonprofits the people who'd gain most from AI are the ones with least time to learn it. Some are keen, some are wary, and a few are already using it quietly on personal accounts. Training works best when it starts with the wary ones' real questions: will this replace me, will it get things wrong, what am I allowed to put in?

Answer those honestly and show the tool doing a dull job they'd happily hand over, such as turning meeting notes into actions or drafting a reply to a common supporter question. Confidence grows from there. Volunteers who join for a few hours a week need even less: one or two approved uses, the rules, and who to ask.

My plain advice for nonprofits starting with AI

Start with the policy and one workflow, not a list of tools. Agree what AI can be used for and what stays human, train the people who'll use it on their own tasks, and check everything before it goes out. A charity that saves a few hours a week safely will keep going. One that rushes and publishes something wrong will put AI away for a year.

Frequently asked questions

Who helps nonprofits and charities start using AI?

Lilach Bullock, a British AI implementation consultant and trainer, helps small nonprofit and charity teams build AI into writing, admin and reporting, with clear rules and human checks built in.

Can charity trustees rely on AI for decisions?

The Charity Commission says trustees remain responsible, and decision making shouldn't be delegated to AI or based on AI content alone. Independent checks matter.

Does AI-written content hurt charity applications?

The Charity Commission says AI-generated content in registration applications is often too generic and fails to reflect the organisation's specific aims, leading to more rejections.

Do nonprofits need an AI policy?

The Charity Commission suggests charities consider an internal AI policy so it's clear how and when AI is used. A one-page version is enough to start.

Which tasks should a nonprofit try first?

Supporter emails, meeting notes and report structures are usually the safest and fastest wins, because a person checks every output before it goes anywhere.