Your charity,
in AI answers.

Say someone in Gosford types 'where can I get free food this week'. Google's guide calls what happens next query fan-out: the AI runs several related searches at once, such as opening hours, whether a referral is needed, and what to bring, then writes one answer with links to the pages that back it up. Google's AI Mode has been live in Australia since October 2025.

ChatGPT works in a similar way when it searches the web for current information. OpenAI says sites that block its search crawler, OAI-SearchBot, won't be shown in ChatGPT's search answers. Some website security settings can block it without anyone choosing to, so it's worth checking.

The point for a charity: the answer is only as good as the pages it can find. If your opening hours live in a Facebook post from March and your eligibility rules sit in a PDF, the AI has little to work with.

The ACNC publishes register details as open data. Its notes say the dataset includes each charity's website 'where supplied', business address, the states it works in and the groups it helps, and that the information comes largely from charities themselves. A blank field usually means the charity never filled it in.

So the register can be out of date in exactly the way you'd least want: an old website address, a closed office, a missing state. Some changes, such as your legal name and Responsible People, have a legal deadline to report: 28 days for medium and large charities, 60 for small ones. Your website address and the groups you help have no such deadline, which is why they drift.

People check it too. The ACNC tells donors to look up a charity on the register before giving, so an accurate entry helps with humans and machines at once.

The quickest way to know where you stand is to ask. Do it for all three of your audiences, not just donors: the person looking for help is usually the most important search, and the one most charities never test.

Whole Whale, a US agency, ran 120 donor-style questions across 12 causes through several AI models. Answers about cancer named only 30 different charities in total, against 127 for the environment and for poverty. In a crowded cause, the same few big names keep coming back.

An October 2026 opinion piece in the Stanford Social Innovation Review, drawing on that study, argues that charities with steady press and consistent profiles show up first, and that conflicting information pushes AI tools toward larger, better-documented groups. The authors say smaller groups often stay hidden unless the donor knows what to ask for.

Whole Whale is a US agency, but the lesson travels. A small charity can't outshout a national one. It can be the most specific, checkable answer for one place: 'free financial counselling in Tuggerah, no referral needed' is something a national charity's page usually can't say.

AI answers can be confidently wrong. When a UK charity tested Google's Bard on race equality funding in the north-east of England, only one of the six funding programmes it recommended actually existed. That was 2023 and the tools have changed, but the habit of checking hasn't gone out of date.

For a charity, a wrong answer has a real cost. A person told you're open on Saturdays turns up to a locked door. So if a tool gets a detail wrong, find where it probably came from: an old page on your own site, your register entry, your Google Business Profile or a directory listing you forgot about. Fix the source and make the rest match.

Google is unusually clear here. Its guide says you don't need an llms.txt file, special schema markup, or to rewrite pages in small chunks for AI. Pages need to be indexed and eligible to show with a snippet. Structured data still helps with other search features, so keep it if you have it.

It also warns against making a separate page for every way a question can be asked. When it's done mainly to rank, that counts as scaled content abuse. One good page per service, written for the people who use it, does the job.