Dale Nirvani Pfeifer
← All insights

November 21, 2026 · 6 min read

Should AI tools be allowed to read your nonprofit's website? The pros and cons.

Many nonprofit websites turn AI tools away without anyone deciding to. Here is what you gain and what you risk by letting them in, and a middle path most organizations can live with.

When we run the Nonprofit AI Visibility Check, some websites turn our checker away. The site loads fine for people, but our automated visit gets an error. Often nobody at the organization knows. A security service, a website plugin or a web host made the choice for them.

Our checker being turned away doesn't matter much on its own. What matters is whether the same setting also turns away the AI tools donors use. From the outside, we can't always tell. Some security settings block anything unfamiliar but let the big AI search tools in; others block those too. Two clues help. Your robots.txt file says which bots are welcome. And if ChatGPT or Perplexity cite pages from your site in their answers, they can clearly read it.

When AI tools can't read your site, it shows. A donor asks "Who runs this charity?" or "Are gifts to them tax-deductible?", and the assistant answers from whatever else it finds: an old press release, a directory listing, a profile from three years ago. In one recent check, an assistant told donors it couldn't confirm who led an organization, while the organization's own About page named its CEO with a photo. The assistant apparently couldn't read that page.

So should you let AI tools in? It's a real choice, with real trade-offs. Here they are.

First, there are two kinds of AI visitors

  • Training crawlers collect web pages to train future AI models. Examples: GPTBot (OpenAI), ClaudeBot (Anthropic), and CCBot (Common Crawl, whose archive is widely used for AI training). Blocking them keeps your content out of future training. It doesn't stop assistants from finding you when a donor asks.
  • Search and answer tools read your pages at the moment a donor asks a question, so the answer can be current. Examples: OAI-SearchBot and ChatGPT-User (ChatGPT), Claude-SearchBot and Claude-User (Claude), PerplexityBot and Perplexity-User (Perplexity).

You can allow one kind and block the other. For most nonprofits, that's the answer.

Google works differently. Its AI answers come from the same crawler as ordinary Google Search (Googlebot), so you can't block Google's AI answers without dropping out of Google Search too. Google offers a separate setting, Google-Extended, that limits some uses of your content for its Gemini models without affecting Google Search. It also controls whether the Gemini app can use your pages when it answers, so blocking it has a cost.

The case for letting AI search tools in

  • Donors get your facts, in your words. Your EIN, your current leader, your real donate link, what you actually do. Without your site, assistants piece you together from other people's pages, and those are often out of date.
  • You fix mistakes at the source. When an assistant gets something wrong, the fastest fix is usually a clear page on your own site. That only works if the assistant can read it.
  • It answers the trust questions. "Is this a legitimate charity?" is one of the first things donors ask. An assistant that can read your financials page, your board list and your tax-deductible wording can answer it. One that can't often tells donors to "check before giving".
  • It costs nothing. No new pages, no fees. It's a setting.

The case for caution

  • Your content gets used without a visit. When an assistant answers from your page, the donor may never come to your site. You lose the chance to show them your donate button, your story and your newsletter sign-up. Assistants increasingly link their sources, but fewer people click than with a normal search result.
  • Training is a separate decision, and a fair one to refuse. Some organizations don't want their writing, their program stories or their clients' stories used to train commercial AI. That's a reasonable position. You can block the training crawlers and still let the search tools in.
  • Sensitive pages need protecting anyway. If your site has client stories, photos of people you serve, staff directories or anything you wouldn't want quoted out of context, that's worth reviewing whatever you decide about AI. Pages that shouldn't be public belong behind a login, not just behind a bot setting.
  • Some bots are heavy visitors. Aggressive crawlers can slow a small site on inexpensive hosting or add bandwidth costs. The usual fix is a limit on how fast any one visitor can load pages, not a full block.
  • Not every bot follows the rules. Your robots.txt file is a request, not a lock. The major companies say they honor it; not everyone does. Blocking doesn't guarantee your content stays out of AI, and allowing doesn't guarantee you get quoted.
  • You can't control what the assistant says. Letting it read your site makes accurate answers more likely, not certain.

A middle path most nonprofits can live with

  1. Allow the search and answer tools listed above, plus Googlebot and Bingbot.
  2. Block the training crawlers if you prefer: GPTBot, ClaudeBot and CCBot. Think twice about Google-Extended: it also controls whether the Gemini app can use your pages when it answers.
  3. Put anything sensitive behind a login, and review client stories for consent, as you would for any public page.
  4. If load is a problem, limit speed instead of blocking. Your web host or a service such as Cloudflare can do this.

Questions to decide with your team

  • How much of your giving comes from people who look you up before they give? The more it does, the more it matters what AI tells them.
  • Is anything on your site that you wouldn't want quoted by an assistant? Fix that page first, whatever you decide about bots.
  • Do you object to your content training AI models? If so, block training and allow search. You don't have to choose all or nothing.
  • Who looks after your website, and did anyone actually choose the current setting? Often the answer is "a default nobody looked at".

How to find out where you stand

Ask whoever looks after your website to check two places: your security or bot-protection settings (Cloudflare, your web host's firewall, or a security plugin such as Wordfence) and your robots.txt file. Cloudflare blocks some AI crawlers by default on new sites, and its defaults have changed more than once, so a site can be blocking AI without anyone having decided to.

Or run the free Nonprofit AI Visibility Check. If your website turns our checker away and there are signs AI tools can't read it either, your report says so, and gives you a message you can send to your web person asking them to let the search tools in and keep the training crawlers out.

Whatever you decide, decide it on purpose. The worst outcome is a setting nobody chose, quietly deciding what donors hear about you.