Why does AI cite weak or outdated sources about my brand?

An AI answer with citations is built in stages. First, a crawler has to be allowed to fetch your page (access). Then, for a specific question, the engine has to pull your page into its working set (retrieval). Only then does the model decide what to use and cite (generation).

If your page fails the first two stages, it is not in the running. The model still has to answer, so it cites the pages it did retrieve. Those can be an old directory listing, a reseller page, a forum thread, or a thin article that happens to be easy to fetch.

Popular sources also get cited far more often. A 2025 study of AI search citations (arXiv:2507.05301) found that the top 20 most frequent news sources accounted for 67.3% of citations for OpenAI models. That is an association with brand size, not a lever you can pull. It does mean a large, weak page about you can beat a small, accurate one.

The practical reading: when AI cites a bad source, first ask whether your own page could have been cited at all.

Each of these is something you can check and fix. None guarantees a citation. Each one, if broken, can remove you silently.

JavaScript-only content

Vercel's crawler study found that ChatGPT and Claude crawlers fetch JavaScript files but do not execute them, so they cannot read client-side rendered content. Googlebot does render. If your pricing, product names or founding facts appear only after scripts run, many AI crawlers see an empty page and turn to someone else's copy.

robots.txt, per bot

OpenAI documents separate bots. GPTBot governs training. OAI-SearchBot governs ChatGPT search, and sites opted out of it will not be shown in ChatGPT search answers. A blanket AI block meant to stop training can remove you from search answers too.

Firewall and CDN blocks

A robots.txt that says allow is not enough if your firewall returns a 403 or a challenge page to the crawler. OpenAI's help centre asks site owners to allow its crawler and confirm the host or CDN allows its published IP addresses. The AI Crawler Checker requests a page as each AI crawler and shows which ones get turned away.

Index and snippet eligibility

Google states that a page must be indexed and eligible to show with a snippet to appear as a supporting link in AI Overviews or AI Mode. A stray noindex, nosnippet or data-nosnippet on a template can hide exactly the facts you want quoted.

The source shown next to an answer is a weak guide to where the answer came from.

• OpenAI's web search API returns the full list of sources consulted separately from the citations, and notes the sources list is often longer.
• Mechanistic research on how models cite (arXiv:2606.28358) repaired over 90% of missed citations by adjusting internal model components, without changing the sources. Citation output is partly separate from which sources influenced the text.

Two things follow. A wrong fact can come from a page that is not cited. And a correct page of yours may have been used without being credited. When sources conflict, the engine has to choose, which is covered in what AI does when sources disagree about your company.

So do not judge the problem only by the citation list. Judge it by the claim: is the stated fact right or wrong?

Old sources show up for three common reasons.

1. No retrieval at all. Many answers are generated without a live search. Then the model relies on what it learned before its knowledge cutoff. OpenAI's model docs list a Feb 16, 2026 cutoff for the GPT-5.6 family and Aug 31, 2025 for the Instant path. A rebrand or price change after that date will not appear in an ungrounded answer.
2. Old pages are still reachable. If an outdated press release or review is easy to fetch and your current page is not, the old one gets retrieved.
3. Low-quality pages can persist in one index. In OtterlyAI's 2026 experiment, Google deindexed two mass-produced AI blog sites, yet Microsoft Copilot, which grounds on Bing, kept citing them. One engine dropping a page does not mean every engine has.

To tell a stale answer from an invented one, see hallucination vs outdated AI answers. Old coverage has its own dynamics, covered in whether old news can shape AI answers today.

Work from what you control outward.

1. List the facts AI gets wrong. Write down the exact wrong claim and the platform. Price, founding year, product names and ownership are the usual ones.
2. Put each fact in your raw HTML. State it as a plain sentence on a page crawlers can fetch without running JavaScript. A specific fact, like a price or a one-line definition, gives an engine a sentence it can quote.
3. Remove contradictions on your own site. If the pricing page, about page and footer disagree, an engine has no clean answer to pick.
4. Check access per bot. Confirm robots.txt allows OAI-SearchBot, PerplexityBot and Claude-SearchBot, and that your CDN does not block them.
5. Ask cited sites to correct errors. When engines cite other sites instead of yours, being accurately named on those sites is associated with appearing correctly in answers. Contact the editor with the correct fact and a link to your source page. For a hostile page, see what to do about a competitor's comparison page.
6. Recheck by platform. Each engine retrieves differently. The fix plan for AI getting your brand wrong and correcting Perplexity's sources go platform by platform.

Avoid shortcuts. Hidden text or instructions aimed at AI crawlers is manipulation, and research on search agents shows it can sway answers, which is exactly why it is a reputational and policy risk, not a fix.

Will adding schema markup stop AI citing the wrong sources? There is no evidence it will. Google says structured data is not required for its generative AI features, and Ahrefs' matched test of 1,885 pages found AI Overview citations fell 4.6% after schema was added. Existing markup does no harm. The stronger move is putting the correct fact as plain text in HTML that crawlers can fetch without JavaScript.

Does allowing GPTBot help ChatGPT cite my site? No. OpenAI documents GPTBot as a training crawler. ChatGPT search uses OAI-SearchBot, and OpenAI says each setting is independent. If you want to appear in ChatGPT search answers, OAI-SearchBot needs to be allowed in robots.txt and reachable through your firewall or CDN. Blocking GPTBot only signals your content should not be used for training.

Can I make an AI engine drop a bad source? Not directly. No engine offers publishers a control over which third-party pages it cites. What you can do is ask the site owner to correct or remove the error, and make sure your own correct page is reachable and states the fact plainly. Some engines also offer in-product feedback, but none guarantees a change.

Why does one AI get my facts right and another gets them wrong? Each engine uses its own index, crawler and retrieval process, and many answers run without live search. A page that is fine in Google may be blocked for another crawler, or missing from Bing. Check each platform separately, and check Google Search Console and Bing Webmaster Tools separately, since one being healthy tells you nothing about the other.

How do I see which AI engines name the wrong facts? Ask each engine the questions a buyer would ask and record the exact claims and cited URLs. Do it logged out with memory off so results are comparable. The free checker on this site runs up to three buyer questions as a one-time snapshot, which is a quick way to see where to start before checking each platform by hand.