We secured our own brand TLD, .reputationpartners. Your brand should too.See how
  1. Reputation Partners
  2. Journal
  3. Removal

16 min readView as Markdown

Wikipedia Mirror Sites: Can You Remove Them?

Wikipedia mirror sites copy articles under an open license. See what the license allows, what operators accept, and how long copies take to update.

Wikipedia mirror sites can be removed only one site at a time, and only on grounds the license does not cover. Wikipedia text is published under an open license that lets anyone copy it, so a mirror that credits Wikipedia and shares the text under the same terms is usually within its rights, even after the original article is corrected or deleted.

That is why the order of work matters. You fix or remove the Wikipedia article first, because every copy descends from it. Then you wait for copies to refresh, and only then deal with the ones that stay: those that break the license, those that add false claims of their own, and those that expose personal data the license never covered. This guide explains the mechanics: what the license permits, what a mirror operator or a hosting company will accept, how long copies take to update, and which search engine routes apply to what is left.

The scope is narrow on purpose: copies of Wikipedia content on mirrors, wikis, knowledge sites and scraped pages, not how to correct a Wikipedia article itself. Rules that differ by country are labeled. Everything here is information, not legal advice.

The short version
  • The Wikipedia license lets others copy and republish the text, so a compliant mirror has no duty to take a page down because you ask.
  • Fix or remove the Wikipedia article first. Copies follow the source on a delay that runs from hours on live mirrors to never on frozen snapshots.
  • Complaints that work target a specific failure: missing attribution, a false copyright claim by the mirror, added defamatory text, or personal data that should not be public.
  • Hosts and infrastructure providers act on what they host and on legal process, not on a request to un-copy an open-license page.
  • Search engine tools can drop or refresh a copy in results without touching the copy itself.

Why the license allows copies of Wikipedia pages

Wikipedia text can be copied by anyone because contributors license it under Creative Commons Attribution-ShareAlike, and that license is written to be irrevocable. In the version 4.0 legal code, Section 2(a)(1) grants "a worldwide, royalty-free, non-sublicensable, non-exclusive, irrevocable license" (Creative Commons, CC BY-SA 4.0 legal code, read for this guide). Section 6(a) ties the term to the length of the copyright itself.

Related readSpent conviction news article: how to remove it (UK)

Two consequences follow. A person named in an article did not write it and does not hold the copyright in those sentences, and the people who did write them cannot simply pull the license back, because the grant is irrevocable on its face. The Wikimedia Terms of Use page could not be read for this guide, so the point rests on the license text alone. Check the Terms of Use before relying on how Wikimedia words the contributor side.

What the license does not give a mirror

The license is a permission, not a blank check. It does not let a mirror claim ownership of the text, it does not cover text the mirror added itself, and it says nothing about whether private data was lawful to publish in the first place. The routes that work against compliant copies rest on those gaps.

What a mirror must do to stay inside the license

A mirror stays compliant by crediting the source, saying the text is licensed, flagging changes and sharing alike. Section 3(a)(1) of CC BY-SA 4.0 requires the reuser to keep the attribution notices supplied by the licensor, to indicate whether the material was modified, and to say the material is licensed under the license with its text or a link. Section 3(b) requires that adapted material be shared under the same license elements or a compatible one, and bars extra terms that restrict what the license gives.

Wikipedia keeps a directory of mirrors and forks, with a note on each site. The entries read for this guide record the failures editors look for: a live mirror with "no mention of the GFDL" (the license Wikipedia used before CC BY-SA), a live scrape that is "misattributed", and several sites that reserve all rights over reused text. The introduction page of the directory could not be read, so this guide does not list its criteria.

Related readRemove Your Name From Justia, Casetext and Case Sites

What happens when a mirror gets it wrong

A breach has a built-in consequence. Section 6(a) says that if the reuser fails to comply, its rights under the license terminate automatically. Section 6(b) reinstates them if the violation is cured within 30 days of the reuser discovering it, or if the licensor expressly reinstates them. A mirror that never credits Wikipedia, or that claims the text as its own, is therefore not protected by the license while the failure lasts.

That does not hand the person described in the article a claim, because the rights belong to the contributors. What it gives you is a precise, neutral complaint to a webmaster or host: the page copies Wikipedia text without the attribution the license requires. The directory records outcomes of that kind: one entry says the contributor whose content was copied wrote to the webmaster and the content was removed.

Irrevocablehow CC BY-SA 4.0 describes the grant to reusers
30 daysthe window to cure a breach and reinstate the license
Monthlyhow often full Wikipedia data dumps run, which feeds many copies

Creative Commons CC BY-SA 4.0 legal code, Sections 2(a)(1) and 6(b); Wikimedia Meta-Wiki, Data dumps page.

The kinds of copies, and which ones update

Copies of Wikipedia differ by how they obtain the text, and that decides how long a correction takes to reach them. The mirror directory shows the range: one entry is described as a "snapshot of all Wikipedia contents (2022)", another "instantly mirrors updates", one "appears to function as a live web scraper", and others are old copies that nobody maintains.

Type of copyHow it gets textUpdate behaviorWhat to expect
Live mirrorPulls from Wikipedia as pages changeFollows edits within hours to daysFixes and deletions usually propagate
Dump-based mirrorLoads a periodic data dumpFollows the dump cycleA lag of weeks is normal
Frozen snapshotOne-time copyNever updatesStays until the operator removes it
Deleted-article archiveCollects pages Wikipedia removedNot tied to the live siteNeeds a direct request
Derivative or rewritten pageCopies, then edits or summarizesIndependent of the sourceTreat as a separate page

The deleted-article row is not theoretical. The directory describes one site as containing a "large selection of deleted Wikipedia articles" and another as including "copies of deleted mainspace articles". Deletion on Wikipedia ends the page on Wikipedia, not every copy made before it.

Why fixing the Wikipedia article comes first

You fix the source first because every compliant copy is a downstream product of it. If the article is corrected, live mirrors and later dumps carry the correction. If you go after copies first, the next refresh can bring the old text back.

Related readSomeone Registered My Name as a Domain: What to Do

Two practical reasons follow. A mirror that re-imports text replaces any local fix with the current Wikipedia version, so a wrong source brings the problem back. And once the article is corrected or deleted, a copy that still shows the old text is stale, which is easier to ask an operator to update than a current one. How to correct the article is its own subject. What matters here is the sequence: the clock for copies starts when the Wikipedia outcome lands.

How long copies take to catch up

Copies catch up on the schedule of their source, and the main shared source has a monthly rhythm. Wikimedia's data dump page says full dumps with historical content run monthly, starting at the beginning of each month, and partial dumps with current page content only also run monthly, starting about two-thirds of the way through. It says the largest wiki, the English Wikipedia, takes about 14 days for a full run. A mirror that loads dumps can therefore trail an edit by a few weeks. The page does not say whether deleted pages' text is included, so assume nothing about that.

The timeline below is our experience of a typical case, not a statistic and not a promise. It assumes the Wikipedia article was corrected or deleted on day 0.

StageUsuallyWhat changesYour action
Day 0Wikipedia outcome landsSource corrected or deletedSave the final state of the page
Days 1 to 7Live mirrors followSome copies change or vanishRe-check the list, do not send requests yet
Weeks 2 to 6Dump-based copies refreshNext import carries the changeRe-check again, then send requests for the rest
Weeks 4 to 12Search results updateDropped pages leave results over timeUse search engine refresh tools
After week 12Frozen and archive copies remainNo automatic changeDirect requests to operators and hosts

Search engines add their own delay. Google's help page says that when a website owner has removed information, it "will eventually be removed from Google Search" through regular updating, and points to its outdated content tool for a faster update.

Related readAsk a Publisher to Anonymize Your Name in an Old Article

What mirror operators and hosts will accept

Mirror operators accept requests that name a specific, checkable failure, and hosts accept requests that fit their own categories. A request to take down a compliant copy because it is unflattering is the weakest request there is. The table sorts the common requests by what each audience tends to accept.

RequestSent toUsually works whenUsually fails when
Update this page to match WikipediaOperatorThe source was corrected and the copy is staleThe operator runs a frozen snapshot
Add the missing attribution or remove the pageOperatorThe copy omits credit or claims ownershipThe copy already credits Wikipedia
Remove text the operator addedOperator, then hostThe added text is false and you can show whyYou cannot separate added text from copied text
Remove personal dataOperator, then hostThe data is private, such as an address or phone numberThe data is public record in the source
Remove after a court orderHost and search engineAn order names the content or the URLThe order does not reach the operator

Hosts and infrastructure providers

A host acts on what it stores, and an infrastructure provider that only passes traffic usually cannot remove anything. Cloudflare's abuse page says it does not host content through its pass-through security and CDN services, that it cannot remove content it does not host, and that it forwards complaints to the website operator and the hosting provider, giving the host the origin IP address. For some complaint categories the reporter can ask it not to forward the complaint to the site operator. On defamation, it says it may block or remove content "determined by appropriate legal process to be defamatory or libelous," and for content it does host it follows the DMCA notice-and-takedown process. That is Cloudflare's own policy as read; other providers differ, and a registrar or a host in another country may apply a different rule.

The host is the party that can switch a page off, so the request goes to the operator first and to the host if the operator ignores it.

Asking a mirror to drop the attribution to you

One quiet tool sits in the license. Section 3(a)(3) of CC BY-SA 4.0 says that if the licensor requests it, the reuser must remove the attribution information "to the extent reasonably practicable." The licensor is the contributor, not the person written about, so this is a right the editors hold, not you. It removes a credit line, not the text, so do not expect it to take a page down.

Related readRemove an Article From Google News: What Works

How to work through copies in order

Work from the source outward, and stop each branch as soon as it resolves. The procedure below suits a person or a brand whose Wikipedia article has been corrected or deleted and who still finds copies.

Clearing downstream copies of a Wikipedia article
  1. Lock the Wikipedia outcomeRecord the final state of the article, the date of the change and the exact sentences that were wrong or removed. Everything else compares against it.
  2. Build a list of copies by searchSearch for distinctive sentences from the old article in quotes, and for your name plus words from the article. Record each URL, the site name and the date seen.
  3. Sort each copy by typeUse the table above: live, dump-based, snapshot, deleted-article archive or derivative. The type decides whether to wait or to write.
  4. Wait out the refresh windowGive live mirrors about a week and dump-based mirrors several weeks. Then re-check the list and drop every URL that now matches the source.
  5. Write to the operator with one specific failureName the page, state the failure (stale text, missing attribution, added false text, private data) and attach the evidence. Keep it short and factual.
  6. Escalate to the hostFind the host from the site's public technical records. Send the same facts and the same evidence, and follow the host's own complaint form.
  7. Ask the search engine to drop or refresh the resultUse the refresh tool for stale pages, and the legal or privacy forms where the grounds fit (see the next section).

Search engine routes for the copies that remain

Search engine tools change what a copy shows in results, not what the copy says. For a stale copy, Google's Refresh Outdated Content tool is the first stop. Its help page says it is for people who do not own the page, that it applies when the page or image no longer exists or has changed significantly, and that a page which is gone drops out of results while a changed page loses its snippet until the next crawl. It says plainly that it "only updates Google's search result" and does not remove a page from the web. It does not handle information that is wrong or harmful, and it does not remove information that is still on the live page. So it fixes the stale-snippet problem and nothing more.

For content that is still live and still wrong, the route depends on the ground.

  • Copyright or court order. Google's removal request page lists court order, defamation and copyright among its legal reasons, each with its own form. A court order that names the content is a strong ground. A copyright request fits only where the requester owns the rights, which is rare for a Wikipedia sentence.
  • Private information. On the same page, personal identifiable information such as card numbers or login credentials sits under Google's policy reasons, and a complaint based on a country's privacy law goes down the legal path. The page says a policy report does not replace a legal one, so each is filed separately. General biographical facts are not what the policy route covers.
  • European data protection law. Google's help page for these requests says the requester is usually the affected person, that Google balances public interest against data protection rights, and that a delisting applies only to results for queries related to the requester's name. The content can still appear for other queries. This route is for people in places where European data protection law applies and is not a United States right.

Bing and other engines run their own forms and rules, which this guide did not read: check the engine's own request page before filing. General removal planning sits on our removal service page.

Derivative pages and AI-scraped copies

Derivative pages are the hardest group because they no longer match any source. A scraper that rewrites a Wikipedia biography, an AI-generated profile that summarizes it, or a knowledge site that merges it with other data creates a new text that the license may not govern in the same way. Three questions decide the route: is the text still close to Wikipedia's, does it add false claims, and does it expose private data?

Related readRemove explicit images from Bing and other search engines

A close copy with the same wording is treated like a mirror: check attribution, then ask for an update. A rewrite that adds false statements is closer to defamation or privacy, and the removal grounds are those of the country where you or the operator sit, not those of the license. The statements that matter are the added ones, so record exactly which sentences are not in the source. Where the page is generated by an AI system and no operator can be found, the host and the search engine are the realistic targets, and there is no fixed timetable for either.

Removing one copy does not stop the next from being generated. The durable fix is an accurate, well-supported source, so that the next copy repeats something true.

When to stop chasing copies

Stop when the next request costs more than the copy can do. Two cases justify going further: a copy that adds false or private content, and a copy that ranks above accurate pages for your name. A stale copy on a site with no search visibility is not worth a legal letter, and a letter can draw attention to a page few would find. For copies that will not move, strengthen the accurate pages that rank for your name so the copy loses visibility. Reputation Partners handles that as suppression, using lawful and accurate content only.

If this is happening to you now

If a Wikipedia article about you has been corrected or deleted and copies are still showing, send us the Wikipedia link, the dates, and the list of copies you found. We sort them by type, tell you which are worth a request and which are better left to fade, and say what can and cannot be removed. Start with a private audit.

Questions people ask

Can I make a Wikipedia mirror take down a page about me?

Not simply because you ask. A mirror that credits Wikipedia and shares the text under the same license is within its rights. You can ask it to update a stale copy, add missing attribution, drop text it added itself, or remove private data.

Does deleting my Wikipedia page remove the copies?

No. Deletion ends the page on Wikipedia. Live mirrors usually follow and dump-based mirrors follow after their next import, but frozen snapshots and deleted-article archives may keep the text indefinitely. Search for distinctive sentences from the old article after a few weeks, and sort what you find by type, rather than assuming the copies are gone.

Is a Wikipedia copy a copyright violation?

Usually not. If the copy credits Wikipedia, says the text is licensed, and shares it under the same terms, it follows the license. A copy that omits those steps may have lost the protection of the license for as long as the failure lasts. That is a matter for the contributors who hold the rights, not for the person described.

How long until a corrected Wikipedia article updates on other sites?

Live mirrors can follow within days. Dump-based copies follow their source's cycle, and full Wikipedia dumps run monthly, with the English edition taking about 14 days to build. Search results then take their own time. Frozen snapshots never change on their own, so plan a direct request for those.

Will an AI assistant keep repeating the old Wikipedia text?

It can, for as long as the text exists in the pages and datasets it draws from. Correcting the source and the copies lowers the chance, but no provider publishes a fixed delay for this, and answers can differ between assistants. Check what assistants say after the copies have refreshed, and report remaining errors through each provider's own channel.

Reputation Partners desk

Notes from the desk. Written by the people who handle the cases.

Tell us what Google says. Quietly.

A partner reads every request personally. Every engagement is quoted individually, after an audit.