Deaths today corrections: rumor checks and verified updates

A false report that someone has died is a particular kind of error, and it needs a particular kind of correction. The person the correction is about is alive, is often reading, and is watching a claim about their own death outrun every attempt to catch it. This article is about how such reports are checked on the day they appear, and what a publication owes when it gets one wrong.

No person is named here and no death is reported. Nothing on this page refers to any actual report about anyone.

What to take away

• A same-day death report is checked by tracing it to its origin and asking whether that origin could know. Most cannot.
• The strongest single signal that a report is false is the person's own activity after the report's timestamp. The strongest signal that it is true is a family or institution saying so. Neither is volume.
• A correction of a false death report must reach the places the report reached, name the error plainly, and stay visible. "We regret the error" is not a correction.

The checks a desk runs on a same-day claim

When a report that someone has died arrives on the day, before anything else, the desk asks a short series of questions. None of them is about whether the claim sounds plausible. All of them are about where it came from.

Where did this first appear? Trace every version back until you reach one that cites nothing. That is the origin. Everything after it is a copy. This is the ordinary method of fact-checking rather than anything particular to death reports: find the original assertion, then ask what the person making it was in a position to know.

Could the origin know? A family member, a representative, a funeral home, an institution with direct knowledge: these could. A fan account, an aggregator, an anonymous post, an account that posts death claims routinely: these could not. A report whose origin could not know is unconfirmed no matter how far it has traveled.

Is the origin what it claims to be? Accounts are impersonated, and a statement that appears to come from a representative or a relative may come from nobody of the kind. Check the account's history, its age, its previous content, and whether the person or institution it claims to be has acknowledged it elsewhere.

Does the timestamp make sense? A report that appears before any event it describes could have happened, or that cites a statement that has not been issued, is not a report.

Is there a second, independent origin? Independent means it could not have come from the first. Two outlets carrying the same wire copy are one origin. A family statement and an institutional statement issued separately are two.

Has the person, or anyone with standing, said anything since? This is the check that closes most false reports. A post from the person's own account after the report's timestamp, a representative's denial, an institution saying that it has no such information: any of these ends it.

A desk that runs these checks and gets no confirming answer holds the item. It does not publish, and it does not publish a denial either, because a denial repeats the claim to people who had not seen it.

Why false death reports happen

It helps to know the usual shapes, not to catalog hoaxes but to recognize a pattern when it arrives.

Some are deliberate: an account seeks attention by announcing a death and collects the reaction. Some are mistakes of identity: a real death, attached to the wrong person of the same name. Some are mistakes of time: an old death, or an old report, resurfacing and being read as new. Some are misreadings: a tribute to a living person, a headline about a role or a character, a post about someone else, taken as an announcement. And some are simply the compounding of a rumor that nobody checked.

None of these is rare. The pattern to notice is the same in each: a claim with no origin that could know, spreading on the strength of repetition. Two of the shapes have their own pages: name collisions among performers in actors who died, and the wave of tributes that follows any report in musicians who died. For what happens to a held item over the following days, see deaths this week.

Correcting a false death report

If a publication has published a death that did not happen, the correction is unlike any other, and it needs to be done with that in mind.

Name the error plainly. The correction says that the publication reported a death, that the report was wrong, and that the person is alive. Not "an earlier version contained an error". Standing commitments such as the Associated Press statement of news values and principles undertake to acknowledge errors openly, and openly means legibly. The reader needs to know what was wrong, and the person needs to see it said.

Put it where the error was. At the top of the page that carried it, in the same feed, with the same reach. A correction buried below the original, or published only in a corrections column, reaches a fraction of the people who saw the claim.

Leave it visible. The original report has been copied; the correction is the only thing that a copier, or a search engine, or a reader arriving later, can find that shows the claim was withdrawn. It stays, dated, with the original marked as withdrawn.

Do not repeat the claim in the correction more than necessary. State what was wrong and what is true. Reproducing the original in full gives it another airing.

Tell the person, or their representative, directly. Before the correction is published if possible. They should not learn that a publication has fixed its error from the same feed that told them they were dead.

Chase the copies. The publication cannot correct every site that repeated it, but it can notify the ones it knows about, and it can request that search results and aggregators update. Correcting the source does not correct the copies, and the copies are where most people saw it.

What the harm looks like

It is worth being concrete, because the temptation to treat a false death report as a harmless mix-up is strong among people who have never been the subject of one.

A living person receives condolences addressed to their family. Their family receives them too. Institutions act: accounts are frozen, subscriptions canceled, engagements withdrawn, and each has to be undone by the person proving they are alive, which is harder than it sounds. Search results carry the claim for months. Years later, a stranger says "I thought you had died", and the person explains again.

The correction cannot undo any of this. It can only make it stop getting worse. That is why it has to be prompt, plain, prominent and permanent.

What this site does

We publish no same-day death reports, for the reasons set out on the parent page on deaths today. Where an entry on any page here is found to be wrong, the correction is dated, visible, and stays. Where an entry should never have been published, it is withdrawn with a note that says so. Nothing on this page refers to any actual report.

Common questions

A death is being reported everywhere. Is that enough to believe it?

No. Trace it to its origin and ask whether the origin could know. Widespread repetition of one unsourced post is one unsourced post.

The person's own account has posted since the report. Does that settle it?

Almost always. Accounts can be run by others, so it is not absolute, but it is the strongest single signal available on the day, and it is enough to hold the item.

A publication corrected a false death report by editing the headline. Is that a correction?

Not an adequate one. A correction names the error, appears where the error appeared, and stays visible. Silent editing leaves every copy of the original standing with no trace.

Should a publication that never repeated a false report publish a denial?

Usually not. A denial spreads the claim to people who had not seen it. Where the person or their representative has denied it publicly, that denial can be pointed to.

Why not publish a report with a note saying it is unconfirmed?

Because "unconfirmed" is dropped in the first copy, and the publication's name travels with the claim. An unconfirmed death report is not published at all.