Fact-Checking Policy

Our Core Rule

1. Pre-Publication Checklist

Before an article goes live, we check that:

• Names and titles are spelled correctly and designations are current.
• Dates and places are correct, and the timeline is consistent.
• Figures and statistics match official data, and units, currency, and scale are right.
• Quotes are accurate and presented in their original context.
• Every key claim has a source that a reader could reasonably trace.
• The headline matches the body and does not overstate what the reporting supports.
• Images and video are of the actual event being reported, not lookalike or older footage.
• Legal risk has been considered, including defamation, contempt, privacy, and reporting restrictions.

2. How We Weigh Evidence

Not all information carries equal weight. We rank evidence as follows:

• Highest confidence. Official documents, court orders and judgments, government notifications, regulatory filings, and official statistics.
• High confidence. Named spokespeople speaking on the record, official press releases, recorded interviews, and live official briefings.
• Medium confidence. Eyewitness accounts once corroborated, and expert comment used for interpretation rather than as proof of an event.
• Low confidence. Social media posts, forwarded messages, anonymous claims, and unverified reports circulating elsewhere. These are never used alone as the basis for a factual claim.

Where the best available evidence is low confidence, we either hold the story or clearly attribute the claim and state that we could not verify it.

3. When We Use Anonymous Sources

Anonymity is an exception, not a convenience. We grant it only when all of the following apply:

• There is a clear public interest in the information.
• The source faces a real risk of retaliation, legal jeopardy, or harm if named.
• An editor knows the source's identity and is satisfied they are in a position to know.
• The information is corroborated by a document or a second independent source.
• The article states why anonymity was granted, in general terms.

We do not allow anonymity to be used to make personal attacks. We honour undertakings of confidentiality.

4. Verifying Images, Video, and Viral Claims

• Reverse image search to establish whether the visual has appeared earlier in another context.
• Trace the original uploader and, where possible, contact them for the circumstances of capture.
• Check geolocation clues such as signage, language, vehicle registrations, architecture, and weather against the claimed location.
• Look for manipulation and synthetic media, including inconsistent shadows and reflections, warped text, mismatched audio, and other signs of editing or AI generation.
• Check context: an authentic image can still be presented misleadingly if it is from a different event or date.

When we report on a false or misleading viral claim, we lead with the verified position and avoid repeating the false claim in a way that reinforces it.

5. How We Rate Claims

Where we publish a dedicated verification of a specific claim, we use this scale:

• True — supported by reliable evidence.
• Mostly True — substantially accurate, with minor inaccuracy or missing nuance.
• Misleading — technically accurate in part, but framed or presented so as to mislead.
• False — contradicted by reliable evidence.
• Unverified — we could not establish the claim either way with available evidence.

With each rating we state the sources we relied on and the method we used, so readers can follow our reasoning and reach their own judgement.

6. Developing and Breaking Stories

• Fast-moving stories are labelled as developing so readers know the picture may change.
• Casualty and damage figures are taken from official sources only, and are attributed with the time they were confirmed.
• As facts are confirmed, we update the article and its updated date rather than leaving stale information in place.
• Where an early report turns out to be wrong, we log and disclose it under the Corrections Policy.

7. Limits of Automated and AI Tools

We may use software to help with transcription, translation, or searching large documents. Such output is a starting point only. AI output is never treated as proof, and a member of our team verifies every fact against a real source before publication. Any AI-generated imagery is labelled.

8. Report an Error or Request a Fact-Check

If you believe we have published something inaccurate, or you would like us to verify a circulating claim, contact us with as much detail as possible. Please include the article link or a link to the claim, the specific statement in question, and any evidence you can share.

9. Changes to This Policy

We review this policy periodically and update it as our verification practices develop. Substantive changes are reflected in the “last updated” date at the top of this page.