Methodology

This page explains every step a claim goes through before we publish a fact check. The same process applies to every fact check.

1. What we check

We check statements of fact about Saudi Arabia that are circulating widely: on X, YouTube and other platforms, in news media, and in statements by public figures and institutions, in any of the languages we monitor.

We do not rate:

• opinions, values or predictions;
• religious rulings, or questions of faith;
• legal interpretations that only a court can settle;
• satire that is clearly presented as satire (unless it is being shared as fact);
• claims about private individuals that have no public interest.

2. How we select claims

• Reach. How widely the claim is circulating, or is likely to circulate: views, shares, number of accounts and languages, and pick-up by media or prominent accounts.
• Importance. Whether the claim could affect people's welfare or well-being, for example their safety, health, legal rights, livelihoods, travel, or how communities are treated.
• Verifiability. Whether reliable evidence exists, or can be found, to test the claim.

Who made a claim, and which position it appears to support, is not a selection criterion. Claims made by governments and officials, by media outlets and by individual accounts are all eligible for checking on the same basis.

To make sure our work is not unduly concentrated on any one source, outlet, community or viewpoint, the Editor-in-Chief keeps a log of the claims we considered and why each was or was not selected, and reviews it every month against what was actually circulating. Readers can suggest claims; suggestions are assessed against the same criteria.

3. How we research

1. Find the original. We trace the claim to its earliest available version and record it, with an archived copy.
2. Contact the claimant where possible. When a claim comes from an identifiable person or institution, we ask them for the evidence behind it, unless that is not possible or would put someone at risk.
3. Use the best primary sources. Original documents, official data, court records, on-the-record statements, original images and video, satellite and geolocation evidence. When we must rely on a secondary source, we say why.
4. Use more than one source. We check every key element of a claim against more than one named source, unless only one relevant source exists, in which case we say so.
5. Treat statements by involved parties as their account. Statements by a government, a company, an armed group or any other party involved in an event are reported as that party's account. We name the source and look for independent corroboration before relying on it, except for a clear official denial or a published official act (see section 4).
6. Check data before using it. We look at how figures and studies were produced before relying on them.
7. Look at all the evidence. Each fact check sets out the relevant evidence that appears to support the claim as well as the evidence that appears to undermine it.
8. Do not guess at motives. We assess what was said, not why someone said it.
9. Protect sources at risk. If naming a source would endanger them, we explain why the source is not named.

4. Our ratings

Each fact check carries one of five ratings. The rating applies to the claim as it circulated, not to the person or outlet that made it.

• True Confirmed by an official source, a document, or two independent reliable sources.
• Missing context True information with context removed that changes its meaning, such as a real number cut from its context.
• Misleading Real material or a real event placed in the wrong place, time or context, or with its meaning distorted. Example: a real clip from elsewhere presented as being in Saudi Arabia.
• False The claim is fabricated, or its material is fabricated, manipulated or AI-generated, or it is contradicted by a clear official denial or decisive evidence.
• Unproven No evidence proves or disproves the claim. This includes cases where the party concerned is silent or says it is investigating, and anything Saudi Arabia has not disclosed.

• We rate only the factual elements of a claim. When a claim has several parts, we rate the central one and explain the others in the text.
• Real material placed in the wrong place, time or context is always rated Misleading, even when it is explicitly attributed to the event. A claim is rated False when it is invented, or when its material is fabricated, manipulated or AI-generated.
• A clear official denial, or a published official act that contradicts the claim (such as a decision in the official gazette), is enough to rate it False. Two denials from the country concerned are enough to rate it Misleading or False, depending on the claim.
• A statement from a non-official involved party, such as a company or an armed group, is not on its own enough to rate a claim True or False; we present it as that party's account.
• When the party concerned is silent or says it is investigating, and for anything Saudi Arabia has not disclosed, the claim is rated Unproven unless decisive independent evidence exists.
• We state facts as a firm yes or no, without "resembles" or "appears". When we say material is not from Saudi Arabia, we cite the visible evidence and do not name the other country.
• The same evidence threshold applies to every claim.

5. How we edit and publish

• Two-person rule. Every fact check is researched and drafted by one person and reviewed by a second before publication. The Editor-in-Chief has final responsibility.
• Language versions. Every language version carries the same rating and the same sources. Translations are checked before publication by: Coming soon
• Short formats. Posts on X summarise what our sources establish and link to the full fact check once published.
• Software and AI tools. We use software tools, including AI tools, to help monitor platforms, translate, search and draft. Tools do not decide ratings. A human editor checks every fact, source and translation and approves every fact check before it is published.

6. After publication

If new evidence emerges, we update the fact check, change the rating if needed, and record the change in the update log. Mistakes are handled under our Corrections policy.

Last updated: 21 October 2026.