Home › How to verify war footage before you share it
How to verify war footage before you share it
In a conflict, the first hours belong to whoever is fastest, and speed is exactly what makes recycled footage work.
Fact-Check · last reviewed 2026-08-24
The short answerTo verify war footage, establish where and when it was filmed before you consider what it shows. Find the earliest upload, geolocate one fixed detail against satellite imagery, and check the claim against an independent record. Most misleading conflict footage is authentic video from another place or another year, so a detector will never catch it.
The pattern to expect
The dominant form of conflict misinformation is not fabrication. It is real footage, correctly filmed, moved in time or space: a strike from another year, a drone clip from a different border, a training exercise labelled as combat. Nothing in the frame is false, so every visual test passes.
This matters more here than anywhere else. Sharing a mislabelled strike can put a real neighbourhood on a target list in someone's mind, and corrections never travel as far as the original.
If you are not certain, the correct action is not to share with a caveat. It is not to share.
The order that works
- Earliest upload firstReverse image search several keyframes, not one. Conflict footage is reposted within minutes, and the oldest copy usually carries the original language and location.
- Geolocate one fixed thingA minaret, a bridge pylon, a factory chimney, a road junction. Fixed structures beat vehicles and uniforms, because they can be matched against satellite imagery and street-level photos.
- Establish the season and time of dayFoliage, snow, shadow direction and length. A summer treeline in a winter claim ends the discussion.
- Read the audioLanguage, dialect, sirens, the specific rhythm of the weapon. Audio is edited less often than pictures and gives away reuse.
- Find the independent recordA strike of any size leaves traces: local reporting, official statements, monitoring organisations. If a major event has exactly one source, treat that as the finding.
Where automation helps and where it must stop
| Task | Can it be automated | Why |
|---|
| Gathering sources on the claim | Yes | Retrieval is mechanical, and it reaches primary documents in 23 of 24 measured cases |
| Checking a quote appears in the cited source | Yes | String matching against the fetched page, published at 73.1 percent verbatim across 633 citations |
| Comparing event date with post date | Yes | Two timestamps, and it catches the most common hoax |
| Geolocating a frame | Partly | Models suggest, humans confirm against imagery |
| Deciding whether a strike was lawful | No | Not a factual lookup, and no tool should pretend otherwise |
| Judging whether the poster meant to mislead | No | Intent is not a fact you can retrieve |
Reading a result honestly
A truth score on a conflict claim is not a verdict on a war. It answers one question: is this specific assertion supported by the evidence found. A high score on a narrow claim can still sit inside a misleading frame, and the framing is where most conflict misinformation lives.
That is why our results show counter-positions rather than a single line, and why sources are never ranked by origin. Across 24 measured claims, 63 percent of everything returned was a primary document, legacy media was 14 percent, and other fact-checkers were 1 percent, a single hit in the entire run. Non-English roots were part of the test on purpose, including Russian authorities and courts outside Europe, because a conflict story told only in English is a story with a hole in it.
Questions
What is the most common type of fake war video?
Real footage from another time or place with a new caption. Because the video itself is authentic, visual analysis and detectors pass it, which is why location and date checks come first.
How do I geolocate a video without special tools?
Pick one fixed structure, a tower, bridge, stadium or unusual roofline, and match it against satellite and street-level imagery around the claimed area. Fixed structures are reliable, vehicles and people are not.
Can I trust footage published by an official military account?
Treat it as one interested source among others. Official accounts publish selectively, and selection is a form of framing even when every frame is genuine. The same standard applies to all sides.
Does wyper verify the video itself?
It verifies the claim around it. It gathers sources, checks that quoted sentences appear in them, and reports the gap between the event date and the post date. It does not run a forensic classifier on the pixels, and it never rules on intent.
What if I already shared something that turned out to be false?
Reply to your own post with the correction and the original date rather than deleting silently. A visible correction travels the same path the mistake did, while a deletion only removes the evidence from your own view.
Check a claim and look at the evidence yourself
Free: 5 checks a day, no account needed. Every statement tells you whether a quote backs it.
Read on
How to check if a video is a deepfakeOrigin, chronology, claim, motive. In that order.
Ask any fact-checker for their error rateOur retrieval, quote verification and model disagreement, measured and published.
wyper