Out-of-context photos & videos

How it works

The video is real. The explosion happened, the crowd was there, the soldier did say that. Only the caption is false: wrong year, wrong country, wrong war. That is enough. Nobody fakes pixels when a new sentence does the job for free.

This is the most common form of visual disinformation, because it is the cheapest. Every new crisis is met within hours by old footage: the 2020 port explosion in Beirut recaptioned as a strike in Ukraine, air-show jets presented as dogfights, a stadium crowd in one country sold as a protest in another. In 2022 the makers of the war game Arma 3 had to publish a statement because clips from their game kept circulating as real combat footage from Ukraine, complete with news-channel logos and a breathless caption.

The trick exploits a simple fact: a picture feels like proof. Text can lie, but “I saw it with my own eyes” settles the matter, even when what you saw happened somewhere else entirely. The caption tells your brain what it is looking at before you have time to ask.

Who does this? Everyone, which is the point. State channels do it on purpose, activists do it in good faith, bored accounts do it for clicks. The footage changes sides as easily as the caption.

Example

@frontline_now, 4h [video: thousands filling a boulevard at night, flares, chanting] THIS is what the regime does not want you to see. Last night in the capital. Share before it gets deleted.

@kata_checks, 2h Reverse image search: this is a football celebration in Buenos Aires, December 2022. Same flares, same balcony on the left. The protest may have happened, but this clip is not it.

Notice what the first post gives you: emotion, urgency, an enemy that “deletes things”, and no date, no street name, no source. The second post gives you something you can check.

Why it works on you

• Seeing feels like knowing. Your brain treats video as witnessed, not told. The caption rides in on that trust. The footage is honest; the sentence above it is doing all the lying.
• It arrives with a feeling attached. Outrage, fear and pride all switch off checking (see emotional hijacking). By the time you have calmed down, you have already shared it.
• It fits what you expected. If you already believe the regime beats protesters, a clip of a beaten protester needs no inspection. That is confirmation bias doing the poster’s work for free.

What to do

When someone sends it to you: ask the three boring questions: when, where, who filmed it. If the post does not say, that is already an answer. Then grab one frame and run a reverse image search. The earliest copy you find usually carries the real story, and the whole thing takes about a minute.

When you catch yourself about to share: look for clues inside the frame. Season, weather, number plates, shop signs, the language on posters, the uniforms. A winter coat in a “yesterday” clip during a heatwave is a clue. So is a video with no sound, or sound that does not match the picture.

In a comment section: do not shout “fake”. The footage is often real, and “fake” lets the poster win by proving the explosion happened. Say what it actually shows and link the original. “This is Beirut, 2020, here is the source” ends the discussion more politely and more completely.

Don’t confuse it with

• Illustrative images in news articles, where a stock photo of a hospital corridor is labelled as such. That is lazy, but it is labelled; the trick is the missing or false label.
• Old footage shared as context (“this is what the same square looked like in 2006”) are fine when the date is stated. The problem begins when “then” is passed off as “now”.