Is this video edited or out of context? How to check

Deepfakes get the headlines, but most misleading videos are made without any AI at all. Fact-checkers call them cheapfakes: real footage that has been slowed down, cut, re-captioned or moved to a different time and place. They're cheap to make and they fool people every day. Here's how to spot the common tricks.

The six common tricks

1. Old footage, new caption. The most common of all: a real flood, riot or accident from years ago, shared as if it happened today.
2. Wrong place. Real footage from another city or country, captioned as local news.
3. Selective cuts. A speech trimmed so a sentence means the opposite, or the part that explains what happened is removed.
4. Speed changes. Slowing a clip down to make someone look drunk or confused; speeding it up to make a crowd look violent.
5. Swapped audio. Real video with different sound: chanting, gunshots or a voice added on top.
6. Cropping and flipping. Cutting out what's around the action, or mirroring the video so reverse search can't find the original.

How to check each one

• Find the longer version. Search for the event and look for the full speech or the uncut video from a news outlet or the original uploader. Context removed by a cut usually reappears in the full version.
• Listen to the speed. Slowed voices sound deeper and slurred; sped-up ones sound high and rushed. Watch people walking in the background: do they move naturally?
• Look for jump cuts. A sudden change in someone's position, the light or the background mid-sentence means something was removed.
• Check the sound against the picture. Do lips match words? Does crowd noise match how many people you can see?
• Read the text in the frame. Backwards letters on signs or clothing mean the video was mirrored.
• Reverse-search a few frames to find the earliest copy, as in our guide to free fact-check tools.

Cheapfake or deepfake?

A deepfake is made or changed by AI: generated footage, a swapped face, a cloned voice. A cheapfake uses ordinary editing on real footage. The difference matters because they need different checks:

• Our checker looks for signs of AI: how the picture moves over time, individual frames, the voice, watermarks, Content Credentials and labels in the post.
• A real video with a false caption will usually come back "most likely real", because the pictures are real. Our fact-check lookup helps when fact-checkers have already covered the claim; otherwise the context check is yours to do.

So ask two questions about every viral video: is it real footage? And is it what the caption says it is?

When the answer matters, check both: the pixels with a scan, the story with the steps above. Then see how to find when a video was really filmed.