What Happens to Coverage When a Publication Shuts Down?

What actually happens to the archive?

It goes dark, usually quickly and usually without warning. A publication that closes stops being a business, and an archive is a cost centre — hosting, domain renewal, a content management system, someone to keep it running — attached to an entity that no longer exists to pay for it.

There is rarely anyone whose job it is to prevent this. Former owners have no commercial reason to maintain a site generating no revenue, staff have moved on, and the decision to let a domain lapse is made by whoever is winding up the accounts.

The result is that the coverage a company bought, pitched for or earned can vanish entirely, on a timetable set by a publisher's corporate circumstances rather than by anything the subject did.

This is the sharper edge of a finding reported earlier in this series. Pew's link-rot study found 38% of webpages from 2013 unreachable a decade later, and that 9% of all losses occurred because an entire root domain had stopped functioning — taking every page on it at once. Publication closures are what that 9% looks like from the inside.

Ownership changes carry the same risk without the finality of a closure. An acquired title may be folded into a new platform, its archive migrated selectively or not at all, and decisions about which decade of coverage is worth porting are made by people with no stake in any individual article.

What the cases look like

MTV News is the starkest recent example. When Paramount Global wound the operation down, the website was pulled offline and more than twenty years of writing and reporting went with it — apparently deleted outright, surviving only in independent archives where anyone had thought to capture it.

Smaller closures produce the same outcome at smaller scale, and there are many more of them. The Hook, a weekly in Charlottesville, took roughly 22,000 stories offline when it folded. Each of those stories was somebody's coverage, and a good number were presumably cited in media strips and pitch decks that no longer resolve.

Closure is not even required. In 2024 thousands of articles disappeared from the Daily Hampshire Gazette and the Greenfield Recorder in western Massachusetts during a content management system migration — ordinary technical housekeeping at publications that remain in business.

That last category deserves emphasis because it is invisible. Nobody announces a migration, no headline records it, and the loss is discovered only when someone follows an old link. A company whose media strip quietly stopped working has no way of knowing unless it checks.

Local titles account for most of the loss by volume and attract least of the attention. A metropolitan daily closing is a news story; a weekly folding is a line in a trade newsletter, and the archive that disappears with it was often the only sustained record of a town's civic life.

The safety net is being withdrawn

Until recently the answer to all of this was the Internet Archive, which crawls and preserves pages so that a copy survives the original. That route is now closing.

Major publishers including The Guardian, The New York Times, the Financial Times and USA Today are restricting the Archive's access to their journalism. Reporting on the trend has put the number of local outlets doing likewise in the hundreds, though we were unable to retrieve the underlying study and report that figure secondhand.

The reasoning publishers give is coherent. They describe the Archive as a “back door” to their catalogues: an open, crawlable copy of paywalled journalism that exposes them to AI scrapers and lets readers bypass subscriptions. Blocking the front door while leaving the back one open would be pointless.

But the consequence falls on preservation rather than on scrapers. As the coverage puts it, this will “leave significant holes in the public record of the internet” — and the illustration is hard to shake: in thirty years, researchers may be unable to see today's newspaper front pages, despite the Archive continuing to exist.

Whether a given publisher currently permits the Archive is checkable in the same way crawler access is, and worth checking before relying on it.

Why publishers are doing it

Because the archive turned out to be worth money, and giving it away free stopped making sense.

Back catalogues have become, in the phrase used in coverage of this shift, “a hot commodity as data to train AI systems”. Publishers are negotiating licensing deals directly with AI companies, and News Corp's arrangement with OpenAI has been reported at more than $250 million across five years.

An asset that can be licensed for that kind of money is an asset a publisher protects. Every open copy weakens the negotiating position, and the Internet Archive — however benign its purpose — maintains exactly such a copy.

This sits directly alongside our own measurement reported elsewhere in this series, where twelve of fourteen major news domains block Common Crawl and eleven block ClaudeBot. The same publishers appear in both datasets. What looks like two stories — news sites blocking AI, news sites blocking the Archive — is one decision about who gets to hold a copy of the archive, and preservation is collateral.

What this means for coverage bought as evidence

That the asset is more fragile than it appears, and getting more so.

Coverage assembled for a purpose that matures later — an immigration petition, a Wikipedia sourcing challenge, a due-diligence review, a funding round years out — will be read at a distance from publication. Between now and then the publisher may close, migrate, redesign or be acquired, and the preservation route that used to catch those failures is narrowing.

The metadata matters as much as the page. Immigration adjudicators require the title, date and author of published material, and those are exactly the details that vanish with the article. A remembered headline is not evidence.

There is also a mild irony worth noting. The publications whose coverage carries the most weight — the large, credible, paywalled titles — are the ones most aggressively restricting both AI crawlers and the Archive. The more valuable the masthead, the more likely the copy is to be locked down.

What this means for anyone buying placement

Archive it yourself, on the day it publishes. Submit the URL to a public web archive, save a PDF, and record the title, publication, date, author and original address in a place you control. The whole exercise takes a few minutes and is the only part of a placement that cannot be taken away from you.

Check your existing media strip. Some proportion of what is listed there has probably stopped resolving, and you will not have been told. This is a ten-minute audit that most companies have never run.

Do not assume the Internet Archive has you covered. Where a publisher restricts it, there may be no copy at all, and finding that out when the page is already gone is finding out too late.

Media Morpheus sells placement, which is a page on somebody else's site, held at their discretion, on infrastructure they pay for and may one day stop paying for. That is the honest description of the product. The archived copy you make on publication day is the version that will still exist when it matters.

Do the same for coverage you did not pay for. Earned pieces are the hardest to replace and the ones nobody thinks to save, precisely because they felt permanent when they appeared.

Methodology and limitations

This report draws on reported cases of archive loss in the technology and journalism press, on coverage of publishers restricting Internet Archive access, and on the Pew Research Center's 2024 study When Online Content Disappears for the underlying prevalence of domain-level loss. Five limits apply. First, the figure circulating for how many local news outlets are restricting the Internet Archive comes from reporting we could not retrieve in primary form, and is described here as reported rather than verified; the four named national publishers come from coverage we did retrieve. Second, the individual cases are drawn from press reporting rather than from our own verification of what remains online. Third, we conducted no measurement of archive survival rates for this report, and no reliable published series on the fate of archives after closure was located. Fourth, the classification of loss modes is our own ordinal framework, not a measurement of frequency. Fifth, publisher policies on archive access are changing quickly and this describes the position at the date of publication.

Conclusion

Coverage is not a permanent record; it is a page on a company's website, and companies close, migrate and delete. MTV News took two decades of reporting offline in one decision, a single weekly took 22,000 stories with it, and thousands more articles vanished from two newspapers that never closed at all — lost to a software migration nobody announced. What changed recently is the safety net. The Internet Archive caught some of this, and major publishers are now restricting it, describing it as a back door that exposes catalogues they would rather license to AI companies for real money. The reasoning is sound from where they sit. The effect is that the public record thins, and anyone holding coverage as evidence should assume the only reliable copy is the one they made themselves.

Frequently Asked Questions

Sources

1. peer-reviewed Pew Research Center, May 2024. When Online Content Disappears, for domain-level loss rates including the 9% of losses attributable to entire root domains failing.
2. trade press Technology and journalism press reporting on the MTV News archive removal, The Hook's closure, and article loss during CMS migrations at the Daily Hampshire Gazette and Greenfield Recorder.
3. trade press Coverage of publishers restricting Internet Archive access, including named national publishers, publisher rationale, and reported AI licensing values.