What Should I Do When ChatGPT Cites a Scraped Glossary Instead of the Original Page?

Why would ChatGPT cite a copy instead of the original?

ChatGPT's search draws on search indexes and its own crawler, and it cites whichever version it retrieves and ranks highest. A copy can win for several reasons:

• Your page is harder to read. Definitions loaded by JavaScript, hidden in accordions that load on click, or placed below long introductions are hard for AI crawlers, which mostly do not run scripts.
• Your page is blocked. A firewall, CDN rule or robots.txt setting may stop OpenAI's search crawler, OAI-SearchBot, while the scraper's copy is open.
• The copy was indexed first, or ranks better in the search results ChatGPT uses.
• The copy is cleaner. Scrapers sometimes strip navigation and ads, leaving one tidy definition per page.
• Search systems struggle to identify originals. Google says that when many versions of an article exist, identifying the original can be difficult.

How to find out whether AI cites your website or scraper sites instead

1. List your distinctive content: glossary definitions, statistics, examples and phrases that only you wrote.
2. Ask the assistants questions those pages answer, such as "What is [term]?" or "What does [term] mean in [industry]?". Run each several times in ChatGPT, Perplexity, Copilot and Google's AI features.
3. Open every cited URL. Compare the text with yours. Note exact or near-exact copies, and whether they credit you.
4. Search for exact sentences from your pages in quotation marks on Google and Bing to find other copies.
5. Check dates. Your content management system, the Internet Archive's Wayback Machine and your server logs show when your page was published and first crawled.
6. Check your own access. Confirm in server logs that OAI-SearchBot and other search crawlers fetch your glossary pages and receive a normal response.

Plagiarism detection services can scan the web for copies of many pages at once, which helps for large glossaries.

What should you fix on your own page first?

• Put each definition at the top of its section in one or two plain sentences, followed by detail and examples.
• Give important terms their own URL, or at least their own heading and anchor, so a passage stands alone.
• Keep definitions in the page HTML, not in scripts, image files or tabs that load content on click.
• Allow the AI search crawlers you want in robots.txt, firewall and CDN settings.
• Add a self-referencing canonical tag and make sure the page is indexed in Google and Bing.
• Add something copies lack: an example from your field, a diagram with a text explanation, or a note on how the term is used in practice.

Changes that affect citations after redesigns are covered in why ChatGPT citations disappear after a website redesign or content update.

How do you get the copy removed?

1. Contact the site owner. Ask for removal, or for a credit and link if you would accept that. Keep the message short and specific: the URL of the copy, the URL of your original and the date you published.
2. Contact the host. If the owner does not respond, find the hosting company through a WHOIS or IP lookup. In the US, hosts that receive a valid notice under the Digital Millennium Copyright Act generally act quickly to keep their legal protection. Most countries have similar notice procedures.
3. Ask search engines to remove the copy. Google and Bing both accept copyright removal requests for pages that infringe your work. Removal from their results also cuts off many AI assistants that rely on those indexes.
4. Report spam. Google's spam policies treat republishing content from other sites without adding value, or with light rewording, as scraping. You can report such sites through Google's spam report.
5. Keep records of each request and response.

Short definitions can be hard to protect as creative expression, while longer explanations, examples and original wording usually are. For large-scale or commercial copying, take legal advice before sending formal notices.

How do you prevent future scraper citations?

• Publish and get indexed first. Request indexing in Search Console and submit URLs to Bing and other engines through IndexNow as soon as pages go live.
• Name your brand inside the content, such as "In [Brand]'s terminology..." or a short attribution line, so copies carry your name.
• Use absolute internal links. Scrapers that copy pages wholesale then link back to your site, which helps search engines identify the source.
• Limit full-text feeds. Offer summaries in RSS feeds rather than full articles.
• Manage bots. Use firewall or CDN rules to slow or block scrapers that ignore robots.txt, while allowing the search crawlers you want. How to do that is covered in how to use robots.txt and a WAF to allow useful AI bots and block scrapers.
• Monitor regularly. Repeat the detection steps each quarter for your most-cited content.

What if the copy credits you?

Some copies link back and name your brand, like a syndicated version. These are less harmful: assistants citing them may still name you. Decide whether to accept them. If you do, ask for a clear credit and link at the top of the page. If the copy outranks and replaces your page, ask for removal or ask that it be kept out of search indexes. Google's advice for legitimate syndication is that partners block indexing of copied content.

What might an illustrative case look like?

Consider an illustrative payroll software company whose 300-term payroll glossary is widely used. ChatGPT answers questions about payroll terms by citing a content farm that copied the glossary word for word. Checks show the company's glossary loads definitions through a script when a letter is clicked, so OpenAI's crawler sees almost no text. The copy has every definition in plain HTML. The company rebuilds the glossary with each term on its own page, definitions in HTML and examples from payroll practice. It adds the brand name to each entry and submits the new URLs for indexing. It asks the copying site's host to remove the pages and files removal requests with Google and Bing. Within two months, the copies disappear from search results and ChatGPT begins citing the company's own term pages.

What do people ask about scraped content and AI citations?

Can I ask OpenAI to stop citing a scraper site?

There is no general process for choosing which sites ChatGPT cites. Removing the copy from the web and from search engines is more effective.

Does a canonical tag stop scrapers from being cited?

Only if the scraper keeps your canonical tag and the search engine honours it. Many scrapers strip it.

Is it worth pursuing small copies?

Prioritise copies that are cited in AI answers or outrank your pages. Ignore those that nobody sees.

Will blocking AI crawlers stop scrapers?

No. Scrapers often ignore robots.txt. Blocking legitimate AI crawlers only removes your own pages from AI answers.