How to spot misleading photos and videos
Some of the most convincing false posts use nothing fake at all. The photo is real and the video is genuine, but the story attached to them is wrong. An AI detector cannot help with this, because there is nothing artificial to find. This guide shows you what to check instead.
1. What misleading context looks like
The researcher Claire Wardle, in a guide for First Draft, calls this "false context". It means genuine content that has been reframed in a harmful way. These are the forms you are most likely to meet.
- Old shown as new. Full Fact checked a video of a crowd in Sheffield with "13 September 2026" written over it. The footage was from a protest on 4 August 2024.
- A clip cut short. In 2024 a clip of Sadiq Khan calling Labour "antisemitic" spread online. It stopped just before he corrected himself seconds later. He had misspoken.
- The wrong caption. A real picture from one place is said to show somewhere else.
- False subtitles. Full Fact found videos of Vladimir Putin and Recep Tayyip Erdogan with subtitles about warning America over Israel. The footage predated the conflict, and the subtitles did not match what the leaders said.
These posts often appear during disasters, protests and wars, when people want pictures quickly. See the news story guide for checking the wider claim.
2. Write down the claim
Be exact about what the post says. Write down where, when and what. "This is the flood in Valencia last night" gives you three things to check. It also stops you being carried along by how the picture makes you feel.
Any text placed over the image is part of the claim, and it proves nothing. Anyone can type a date or a place on a video.
3. Find the earliest copy
If a picture was online before the event it claims to show, the claim is false. A reverse image search is the quickest way to learn this.
- Follow our guide on how to reverse image search a photo, and use at least two tools.
- In Google Images, select the three dots next to a picture to open "About this image". It can show when Search first saw the image or similar ones, and how news and fact-checking sites describe it.
- For a video, take screenshots of a few clear moments and search each one. The free InVID-WeVerify plugin for Chrome and Firefox can split a video into key frames for you.
Compare the captions on the earliest copies with the claim. Check them against each other as well, because page dates can be wrong and early copies can carry false captions too.
4. Find the full version
A short clip of a speech or interview may leave out the part that changes its meaning.
- Find out where it came from. Broadcasters, parliaments and public bodies often post full recordings.
- Watch at least a minute either side of the moment being shared.
- Look for what was removed. Did the speaker correct themselves, quote someone else or answer a question you cannot hear?
- With photos, look for a wider version. A crop can leave out who else was there.
Editing can go further than cutting. In 2019 Full Fact checked a video of Keir Starmer on Good Morning Britain. It paired the end of a question with earlier footage of him listening, so he looked unable to answer. In fact he replied at once. When the footage has been rearranged like this, the honest verdict is altered, and calling it misleading context would understate it.
5. Check the place and time
The picture itself can contradict the caption. Look for details you can check:
- signs, shop names and number plates, and the language they use;
- landmarks, which you can compare with a map or street-level photos;
- the weather and season, which should fit the reported date and place;
- uniforms, vehicles and clothing;
- photos of the same event from other angles.
A credit to a photographer or news agency is useful, because you can look for the original caption at the source.
6. Check the words
If a video has subtitles or a voiceover in another language, the footage may be real while the words are invented.
- Find the original speech. Full Fact traced the Putin clip to a December 2022 meeting, then compared the subtitles with the Kremlin's transcript.
- Ask someone who speaks the language, or check whether news reports of the speech mention what the subtitles say.
- Check the same for quotes placed on a photo. If a person said something striking, it will usually have been reported.
7. Search for fact-checks
Recycled pictures come round again and again, so one may already have been checked. Search Google Fact Check Explorer with a few words about the image. It can also check whether an uploaded or linked image has appeared in a fact-check. In the UK, look at Full Fact.
Passing an AI check does not make a post true. Content Credentials can show a genuine capture, and the caption can still be false. See what Content Credentials can and cannot tell you.
8. Decide what you can say
- Verified. You found the original from a reliable source, and its time, place and meaning match the claim.
- AI-generated or altered. The picture or sound was changed, or the footage was rearranged to show something that did not happen.
- Misleading context. The content is genuine but comes from another time, place or event, or a cut has changed its meaning.
- Unverified. You could not find the origin. Finding no earlier copy does not prove the picture is new.
Say what you found, including the real date or place if you know it. That is more useful than calling something fake. If you cannot verify it, do not share it as fact.
In the news
Recent reports from RELTY's AI news that show why this check matters.
- 10 November 2026 · DW
Fake and AI-manipulated images spread after the Krakow shooting
DW Fact Check says fake and AI-manipulated images circulated in the hours after the attack and fuelled false stories about it. It examined which material was genuine. The first hours of a breaking story are when pictures need the most care.
- 6 November 2026 · DW
Recycled and AI-made videos in posts about the Iran war
DW Fact Check says posts about the new US attack on Iran include recycled videos as well as AI-generated ones, alongside misleading claims from dubious accounts. Old footage resurfaces quickly when a war is in the news.
- 29 October 2026 · Full Fact
Old AI mushroom cloud video shared with a new false story
A reverse image search led Full Fact to a 2025 AFP check of the same clip, then shared as an Iranian test. This time it was said to show Ukrainian losses. The clip was AI-made, and the search for earlier copies is what linked the two stories.
Sources
The pages this guide relies on, checked on 1 November 2026.
- Understanding information disorder. Claire Wardle, First Draft, September 2020.
- Old video of crowd running through streets of Sheffield shared as if recent. Full Fact, September 2026.
- Clip of Sadiq Khan calling Labour 'antisemitic' has correction edited out. Full Fact, February 2024.
- Fake subtitles show Putin and Erdogan 'warning America' over Israel support. Full Fact, October 2023.
- This footage of Keir Starmer being interviewed on Good Morning Britain was edited before being posted by the Conservative party. Full Fact, November 2019.
- 4 ways to use Search to check facts, images and sources online. Google, April 2024.
- InVID verification plugin. InVID project.
- Fact Check Explorer. Google.
How to check if a viral video is real
How to reverse image search a photo
What to do if someone makes a deepfake of you
How to spot fake reviews and listings made with AI
Browse the RELTY directory of tools
Links to other organisations are for reference only. RELTY has no partnership with, or endorsement from, any of them.
