How Do Publications Handle Corrections, and Can You Request One?

What do the codes actually ask for?

Prompt, visible correction. The Society of Professional Journalists' Code of Ethics pairs an obligation to seek truth and report it with a duty to be accountable, which includes acknowledging mistakes and correcting them quickly, clearly and prominently. (That is the 2014 wording. The revision adopted on 4 October 2026 says “Correct them promptly and prominently.”)

Each of those three adverbs is doing work. Quickly, because an uncorrected error propagates. Clearly, because an ambiguous correction leaves readers unsure what changed. Prominently, because a fix buried where nobody looks does not undo the impression the error created.

This is a genuinely strong commitment, and it is the part of journalistic practice most useful to a company that has been reported inaccurately. Publications that take their standards seriously want to correct errors, because an uncorrected error is a standing liability to their credibility.

It is also the opposite of the dynamic around removal requests, examined elsewhere in this series, where guidance sets a presumption against deletion and warns that a subject's influence should not enter the decision. Correction and removal are different asks with very different odds.

There is a practical reason this obligation is stronger than it looks. A publication's standing rests on being reliable rather than on never erring, and a visible correction demonstrates the first while admitting the second. Editors generally understand this better than the subjects who approach them, which is why a well-evidenced correction request is often met with more cooperation than the requester expects.

How does correction work in the archive?

By amendment rather than replacement, which is the practice journalism's archive guidance recommends.

The guidance contemplates that a publication “will update or correct archived content, including headlines”, while instructing that it “amend the original item to add that update, but not remove the record from the site”.

Alongside that sits the transparency step: add a note saying when the item was updated and what the original headline said. The reader can then see both that a change occurred and what it was.

For a subject this is a better outcome than it first appears. A corrected article with a visible note is a page that now states the accurate position and demonstrates that the publication checked. That reads considerably better to anyone examining it later than an article that quietly disappeared.

There is a version of this that serves a subject better than removal ever would. An article carrying a dated note recording that a figure was corrected demonstrates that the publication checked and that the current text is accurate. Anyone assessing the company later reads a verified page rather than an absence, and an absence invites its own questions.

What makes a correction request succeed?

Specificity and evidence, in that order.

A request that identifies the precise sentence, states what is wrong with it, and supplies the document establishing the correct fact is a request an editor can act on in minutes. One that expresses displeasure with an article's tone gives them nothing to do.

The distinction editors draw is between fact and characterisation. A misstated revenue figure, a wrong date, a misspelled name, an incorrectly attributed quotation — these are correctable. An unflattering but accurate framing is not an error, and asking for it to be changed as though it were will usually end the conversation.

Speed matters too. A correction requested within hours can often be made before the piece has been widely read or syndicated. The same request weeks later has to chase copies that have already propagated, which is one of the reasons the codes emphasise correcting quickly.

Tone on the requester's side matters more than it should. A note that identifies an error politely and supplies the evidence gets handled as routine editorial work. The same error raised with threats is routed to lawyers, where it stops being a correction question and becomes a legal one that nobody at the publication can resolve quickly.

Approach the right person, too. Most publications name a corrections contact or an editor responsible for standards, and a request sent there is handled as routine work rather than forwarded from a general inbox to whoever happens to open it.

What about syndicated copies?

They are the hard part, and the reason a correction at the source is worth pursuing immediately.

Where an article has been picked up, republished or distributed through a wire, correcting the original does not automatically correct the copies. Each host holds its own version, and whether a correction propagates depends on the arrangement between them.

This compounds a problem documented throughout this series. Syndication multiplies reach and multiplies error, and the same mechanism that made a release appear on dozens of domains makes an inaccuracy appear on dozens of domains.

The practical response is to identify the copies early and approach the ones that matter directly, while recognising that some will never be fixed. It is also an argument for accuracy at the point of origin, since a claim in a press release travels further and corrects worse than almost anything else a company publishes.

Wire distribution deserves particular caution for this reason. As documented elsewhere in this series, a release may appear on dozens of endpoints within hours, each holding an independent copy. An error caught after distribution is effectively permanent across most of that footprint, however quickly the original is fixed.

What should a company do about its own errors?

Apply the same standard, which almost no company does.

The codes ask publications to correct quickly, clearly and prominently, and fact-checking's own standards, examined elsewhere in this series, ask signatories to maintain a visible corrections policy and to ensure users see corrected versions where possible.

Most companies have no equivalent. A figure that turns out to be wrong is quietly edited or left in place, and the press release carrying it remains in circulation indefinitely with no note that anything changed.

Adopting the practice costs little and is worth more than it appears. A company that visibly corrects its own published claims has an answer when a journalist or an adjudicator finds a discrepancy, and the answer is considerably better than not having noticed.

A minimal version costs almost nothing: a dated note at the foot of any corrected page saying what changed and when. That single habit puts a company ahead of most of the market on a standard that journalism, fact-checking and platform policy all independently endorse.

The same discipline applies to figures quoted from third parties. A statistic repeated from a vendor study, a trade body or a press release becomes the company's claim once it appears under the company's name, and the obligation to correct it does not transfer back to whoever published it first.

What this means for anyone buying placement

Ask for corrections rather than removals. The codes support the first and set a presumption against the second, and a correction usually produces a better artefact than a deletion would.

Bring evidence and be specific. Identify the sentence, state the error, supply the document. Editors act on that and cannot act on dissatisfaction.

Move fast, especially where syndication is involved, because copies stop being reachable quickly and each one holds its own version.

And get it right before publication. Media Morpheus sells placement and distribution, and a claim that cannot be supported will travel exactly as far as one that can — but only one of them stays correctable. The cheapest correction is the one made in the draft.

Methodology and limitations

This report draws on the Society of Professional Journalists' Code of Ethics for the obligation to correct errors promptly and prominently, and on Online News Association archive guidance for how corrections are handled in published archives, together with the IFCN corrections commitment examined elsewhere in this series. Five limits apply. First, these are voluntary professional codes: they bind nobody, individual publications set their own policies, and many publish none. Second, no data is reported on how often correction requests are made or granted, because no reliable published series was located. Third, the classification of likely outcomes is our own ordinal reading of the guidance, not a measurement. Fourth, nothing here addresses legal remedies for inaccurate publication, which differ by jurisdiction and are outside the scope of this report and not legal advice. Fifth, practice varies widely between well-resourced newsrooms and publications without editorial staff.

Conclusion

Correction is the request journalism is built to grant. Its codes ask for errors to be acknowledged and fixed quickly, clearly and prominently, and its archive guidance describes amending an item and recording what changed rather than removing it. That makes a correction request structurally different from the removal request examined earlier in this series, where the presumption runs the other way and a subject's influence is explicitly not supposed to count. What decides the outcome is whether there is an identifiable factual error and whether the person asking can evidence it. Tone is not an error, syndicated copies rarely all get fixed, and the only correction that reliably works everywhere is the one made before publication.

Frequently Asked Questions

Sources

1. trade body Society of Professional Journalists. SPJ Code of Ethics, for the obligation to acknowledge mistakes and correct them quickly, clearly and prominently.
2. trade body Online News Association. Removing material from your archives, for updating, appending and de-indexing practice.
3. trade body International Fact-Checking Network. Code of Principles, cited for the corrections commitment.