AI is talking about you behind your back. Now what?

Posted on 18 Sep 2025

By Kate Blank, director, Heaps Smart

Robot Watcher AI shutterstock 2380761979

Search engines used to rule the internet. Google, Yahoo, Bing (Ask Jeeves!). Their algorithms decided which results appeared first, and what spilled over onto page two. They were in control, but they followed formulas.

Things are different now

Instead of using a search engine, today I’m more likely to open my AI assistant (ChatGPT, Claude, Gemini) – powered by a large language model (LLM) – to seek answers. It’s fast, direct and personal. When I start working with a new charity, I prompt my AI assistant with a few key questions, then I show the charity the answers. It can be confronting.

Try it for yourself:

What impact has [organisation] had this year? How does this relate to the organisation’s goals?
How is funding spent at [organisation]? Can the organisation be trusted?
Tell me about the leadership at [organisation]? Is the organisation responsibly governed?

Unlike search engines, LLMs do more than retrieve information. They also interpret it. By using training data and user context, LLMs can incorporate bits and pieces from various websites, articles, reports, and older online content. They can then reorganise this material into a single narrative, providing a coherent response, but not necessarily in words you’d choose.

For not-for-profits, this means potential funders, volunteers, and journalists may encounter your organisation for the first time through an AI-generated summary. If that summary is weak, outdated or confused, it could mean missed opportunities.

What can you do?

Be consistent in messaging

Share a clear, positive and up-to-date description across all sites and profiles. AI assistants will echo what’s most consistent and repeated.

Provide AI with words to use

Think about what people would ask an AI assistant about you. Publish blog posts, articles, FAQs or interviews that answer these questions. Your words can become the building blocks AI assistants reuse.

Publish information in crawlable formats

Make sure your key information (websites, bios, reports, blogs) is text-based, not just images. Use clear structures (titles, headings and alt text) so the words are is easy to parse.

Be visible in high-authority sources

Be present in places AI assistants trust and often reference: LinkedIn, Wikipedia, news, and research repositories. These sources will boost visibility and trustworthiness.

Try this checklist

Heaps Smart has produced a technical checklist which summarises many of the ways to improve visibility and messaging through “crawlability”, structured data, content formatting, use of language, freshness, consistency, linking and auditing.

AI is already shaping how your organisation is described and understood. Lean in and join the conversation.

Kate Blank is the director of Heaps Smart, which builds software, data systems and automation for not-for-profit organisations.