What Is reference-url-citation.invalid? How to Recover the Real Source

reference-url-citation.invalid is not a website, a database, or an academic identifier. It is a placeholder address that shows up in text produced by AI chat tools, in the spot where a link to a source should be. The domain ends in .invalid, which is reserved so that it can never point anywhere — the link will not open for you or for anyone else.

Internet standards set aside a few top-level domains that will never be assigned to anyone. RFC 2606 reserves .invalid for names that must be obviously invalid, and RFC 6761 tells DNS resolvers to answer every .invalid lookup with "no such domain". The IANA registry of special-use domain names lists it for the same purpose.

So no one owns reference-url-citation.invalid, no server answers for it, and no archive holds a copy. That is by design: a tool that prints a .invalid address is saying, in effect, "a link belongs here, but I do not have one to give you".

You will usually see the placeholder with a number after it, such as https://reference-url-citation.invalid/0 or /2, and the numbers change between placeholders in the same answer. They work as counters, not as page or article IDs. Searching for /3 will not lead you to a third source, and neither will a DOI, library, or archive lookup.

Seeing the placeholder does not tell you whether a real source was ever behind it. Sometimes a real page was cited and its address was lost on the way into your document. Sometimes there was no source at all. Treat every placeholder as unverified until you have found the actual work.

Work from the text around the placeholder, not from the placeholder itself:

• Go back to the original conversation, if you still have it. A source panel, footnote, or hover card there may show the real address that did not survive copying.
• Read the visible label next to the link. Text like "openai.com" (https://reference-url-citation.invalid/0) names the site even though the address is missing — start your search there.
• Search for the exact title, if there is one, in quotation marks, in a scholarly index as well as a general search engine.
• If there is no title, search for the specific claim the citation was supposed to support — a statistic, a quotation, a finding — and see who actually published it.
• Ask the AI tool for the author, title, and year of the source, then find that work yourself. Do not accept a new link without opening it.

Once you have a candidate, check that it is the right one: the author, title, publication date, and publisher or journal should all belong to the same record, the DOI or URL should open to that record, and the source should actually say what your text claims it says. Our guide on how to check if a citation is real walks through each of those checks.

To check faster, paste the full reference — author, title, year, and venue, not the placeholder link on its own — into the Citation Checker. It looks the reference up in academic databases and shows the matched record, or tells you that nothing matched. For a whole reference list, use the Reference Checker.

If none of the above turns up a real source that says what your text says, remove the citation — and, if nothing else supports it, the claim it was holding up. A statement with no citation is easier to defend than one cited to a source that does not exist. Why AI invents references explains why this happens, and what to do when a citation fails a check covers how to decide between fixing, replacing, and removing.

• Do not delete .invalid and try the rest. reference-url-citation is not a real site under any other ending.
• Do not guess a domain, such as swapping in .com or the name of a well-known publisher, and cite the result without opening it.
• Do not put the placeholder in your bibliography. A reviewer who clicks it will see a link that goes nowhere.
• Do not treat a confident rewrite as a fix. A second AI answer can produce a plausible address for a source that still does not exist.