Do donors really use AI to choose charities?

Some do, but the share is small in the research published so far. A survey of 1,728 donors found that 4.5% use AI to find and research causes. The habit is new, and the figure may change.

Is AI traffic to charity websites large?

Not yet, in the one study we found that measured it. Across 24 nonprofit websites, AI-driven traffic was 2% of total volume from January to October 2025, up about 1,000% on the year before.

Should a small charity change its strategy because of AI?

It should check what the assistants say about it and keep its public facts accurate and consistent. Nothing in the evidence so far justifies dropping other channels.

Does Citewake have data on charity recommendations?

Not yet. Citewake's charity research programme starts in February 2027. Until then, the guides here describe other people's published research and what you can observe yourself.

How donors use AI to choose a charity, and what it…

Some donors now ask an AI assistant where to give, and the assistant answers with a short list or a single recommendation. The published research so far suggests this is still a small share of donors, but a fast-growing one. For small charities the practical question is simple: when someone asks an assistant about your cause or your area, are you in the answer, and is what it says about you correct?

Citewake is an independent research lab. It records what ChatGPT, Claude, Gemini and Perplexity recommend. It has no charity results yet. Its charity research programme starts in February 2027, and the results will be published as Special Edition: Charities (see the research agenda and the categories). Citewake also runs the Citewake Foundation, a free programme for verified charities. This guide covers what others have published, with sources, and says where the evidence is thin.

How many donors use AI to choose where to give?

Fewer than the headlines suggest, so far. Blue State, a communications agency, surveyed 1,728 donors for a report on how brands are discovered in the age of AI. Candid's summary of that report says 4.5% of surveyed donors use AI to find and research causes to support, and that organic search was the third most common way respondents found organisations, behind word of mouth and social media (Candid, December 2025).

That is one survey, taken in 2025. A different donor group, a different country or a later date could give a different figure. Treat 4.5% as a reading from one sample, not a measure of all donors.

How much traffic does AI send to charity websites?

The same Blue State report looked at traffic to 24 nonprofit websites from January to October 2025. According to Candid's summary, AI-driven traffic made up 2% of total volume, which was about 1,000% more than a year earlier. Over the same period, organic search traffic from search engines fell 35% compared with the same months of 2024.

The M+R Benchmarks Study for 2026 describes a related pattern in a larger group of organisations: the share of organic traffic to nonprofit websites fell month by month through 2025, as chat-based search and AI summaries replaced traditional search for many users (M+R, April 2026).

Both sources describe websites in the United States. Neither says why any single charity gained or lost visitors.

What does an AI recommendation look like to a donor?

An assistant tends to give one confident answer rather than a page of links. In an October 2026 article for Stanford Social Innovation Review, Andrew Dunckelman and Dale Nirvani Pfeifer describe a donor who asks where to give to save children's lives. The chat tools they tried pointed to GiveWell's top charities, and ended with a line to the effect of "if I were making one donation, I'd make this one." The authors say the advice is worthwhile, and add that it leaves out smaller groups that evaluators have not rated (SSIR).

That is one example from one article. It shows the shape of the issue and is not a measurement. Measuring it is the purpose of the programme described in why AI may favour large charities.

Do donors who arrive from AI give?

The evidence is mixed and thin. In Candid's summary of the Blue State data, visitors who arrived from AI stayed more than 70% longer than other visitors. Their donation rate was about one-eighth of the average across all traffic sources. Among those who did give, the average gift was $250, against $210 for donors from non-AI search sources.

Those numbers come from 24 websites in 2025. They say AI visitors read carefully and gave less often. They do not say why, and they will not apply to every charity.

What does this mean for a small charity?

Three things follow from what is known.

Find out what assistants say about you. You cannot manage what you have not seen. The guide on how to check whether ChatGPT, Claude, Gemini or Perplexity recommends your charity gives a method.
Keep your public facts accurate and consistent. Candid ran a small test of free AI tools on Candid and 10 other nonprofits. The tools relied mainly on IRS Form 990 data and pointed users to the organisation's Candid profile. Where small organisations filed the short Form 990-N, the tools often treated the thin financial detail as a lack of transparency (Candid, August 2025). Check what your registers and profiles say.
Measure before you spend. AI referrals are small today. The guide on measuring AI referrals and donations explains what you can and cannot attribute.

What should a charity not conclude yet?

Do not conclude that AI is the main way donors will find you. Do not conclude that it is irrelevant either. The honest position is that the channel is small, growing, and poorly measured. Citewake's own analysis of how the four assistants answer buying questions in other fields is in what four AIs did when we asked them 180 buying questions, and the method behind it is on the methodology page.

If you want a quick look at how the assistants describe a website, the Citewake checker is free to use.

How donors use AI to choose a charity, and what it means for small charities