Most false videos I see shared in family WhatsApp groups are not deepfakes. They're real videos with a wrong caption. A flood from 2019 posted as last night's storm. A protest from one country labeled as another. A clip cut ten seconds before the part that changes the meaning.

That's good news, in a way. You don't need special software to catch most of them. You need a routine you can run in five minutes, and the habit of running it before you hit forward. This is the one I use.

Step one: pause before you share

This sounds too simple to count as a step. It's the most useful one.

Viral videos are usually viral because they make people angry, scared or excited. That feeling is what pushes the share button. So when a video gives you that jolt, treat it as a signal to slow down, not to speed up. Nothing bad happens if you share it twenty minutes later, after checking. Something bad can happen if you share a fake to 200 people in a group.

Figure out what the video is actually claiming

Before you check anything, write down (even just in your head) the specific claim. Not "this video about the airport". Something concrete like: "This shows police at Lahore airport yesterday stopping passengers from boarding."

That sentence has parts you can check. Which place? Which date? Who is doing what? A lot of misleading videos fall apart at this step, because when you try to state the claim clearly you notice the video itself doesn't show it. The video shows a crowd and some uniforms. The caption is doing all the work.

Separate what you can see from what you're being told. They're often different.

Check the date and the original uploader

Old video recycled as new is the most common trick by far, in my experience. So find where the video first appeared.

Look at the account that posted it. Is it the original source, or a repost account that collects dramatic clips? Check the upload date on the earliest copy you can find. On YouTube, the date is under the video. On TikTok and X, it's on the post. Search the platform for a few words from the caption and sort by oldest if you can.

Look for clues inside the video too. Weather that doesn't match the season. Shop signs, car number plates, or the language people are speaking. Phones in people's hands (a crowd holding old flip phones is not from last week). Sometimes the clue is obvious once you look for it.

Run a keyframe reverse search

Reverse image search works on videos too, you just do it on stills.

  1. Pause the video at a clear, distinctive moment. A wide shot with buildings or a sign is better than a blurry close-up of a face.
  2. Take a screenshot. Take three or four from different moments.
  3. Upload each one to Google Lens, TinEye, or Bing Visual Search.
  4. Look for older results, especially news articles or videos posted months or years before.

There are browser tools like the InVID-WeVerify plugin that pull keyframes out of a video automatically, which saves time if you do this often. But screenshots on a phone work fine.

If the same frame shows up in a news report from 2021 about a different city, you're done. The video is real footage of something else.

Look for the same event from other angles

Real events with lots of people around get filmed by lots of people. If something big really happened in a public place, there will usually be other videos from other angles within hours, and reporters will pick it up.

So search for the event itself, not the video. Use the place name, the date, and a couple of keywords. Check a few reputable news outlets, ideally ones from that country or city. If nobody else has it, especially for something that would be big news, be careful. It's not proof of a fake (small local events can go unreported), but it's a strong reason to wait.

Then check fact-checking sites. Many countries have them, and the big international ones (AFP Fact Check, Reuters Fact Check, Snopes, Full Fact) often cover viral videos within a day or two. Search the site for a few keywords from the claim. Someone may already have done all this work for you.

Watch for clipped context, edits and swapped audio

Even real, recent footage can lie.

Clipped context. A video that starts in the middle of an argument, or ends right before the reply, can make anyone look bad. If you can find a longer version, watch what happened before and after.

Misleading captions and text overlays. The text on screen is not part of the evidence. Anyone can type anything over a video.

Cuts. Watch for jumps, where a person's position or the background changes suddenly. Two clips stitched together can create a conversation that never happened.

Replaced audio. Sound is easy to swap. A video of a crowd with added chanting, or a politician's speech with a different soundtrack. If lips don't match the words, or the background noise suddenly changes, be suspicious.

AI voice clones. This one is newer and harder. Voice cloning tools can now copy a real person's voice from a short sample. Signs include flat emotion, odd pauses, breathing in the wrong places, or a voice that sounds right but uses words the person wouldn't use. Often the video is just a still photo or old footage with a cloned voice on top. If a famous person is saying something shocking, check whether any real news outlet reported them saying it. If you want more on spotting generated visuals, I wrote about that in how to spot AI-generated images.

A short checklist to save

Here's the whole routine in one place. I keep something like this in my notes app.

  • Did it make me feel something strong? Slow down.
  • What exactly is the claim (place, date, who, what)?
  • Does the video actually show that, or only the caption?
  • Who posted it first, and when?
  • Do clues in the video (weather, signs, language) fit the claim?
  • What does a keyframe reverse search turn up?
  • Are there other angles or news reports of the same event?
  • Has a fact-checking site covered it?
  • Is there a longer version that changes the meaning?
  • Does the audio match the video?

You won't need every step every time. Often the reverse search or the upload date answers it in two minutes.

Where TrueCheck fits

I made TrueCheck because doing the search part by hand is slow, especially on a phone. You paste a TikTok or X link, or share the post to the app. It pulls out the factual claims, searches the web, and gives you a Truth Score with a plain-English summary and links to the sources it found. The first check is free, and after that it uses credits.

The part I care about most is the source links. Please actually open them. A score is a summary of what the app found, and like any automated tool it can miss context or pick up a weak source. I think of TrueCheck as a way to skip the first ten minutes of searching, not a way to skip thinking. Read the sources, see if they really say what the summary says, and then decide.

And if after all of that you still aren't sure, the simplest choice is just not to share it. Nobody in your group will miss one video.