Reverse video search: find where a clip came from

Finding the original upload is the single most useful thing you can do with a suspicious video. It tells you who filmed it, when and where, and it quickly exposes old footage with a new caption. There's no "reverse video search" button that works for every clip, but the method below works for most.

The idea: search images, not video

Search engines index images far better than video. So instead of searching the video, you search a few frames from it. If the footage has been online before, at least one frame usually matches.

Step 1: Pick good keyframes

Not every frame is useful. Choose frames that are:

• Sharp, not mid-movement or blurred.
• Distinctive: a landmark, a sign, an unusual building, a clear face, a logo.
• From different parts of the clip: the start, a key moment and the end.

On a phone, pause and take a screenshot. On a computer, the free InVID-WeVerify browser plugin, built for journalists, can split a video link into keyframes for you.

Step 2: Crop to what matters

Remove captions, emojis, platform logos and black bars added by re-uploaders. They confuse search engines. Cropping to just the distinctive part of the scene often finds matches a full screenshot misses.

Step 3: Search in more than one engine

Each engine has different strengths, so try at least two:

• Google Lens: the best general starting point, and built into Android, the Google app and Google Photos.
• Bing Visual Search: sometimes finds matches Google doesn't.
• Yandex Images: often strong for faces and for footage from Eastern Europe and Central Asia.
• TinEye: lets you sort results by oldest, which is exactly what you want when hunting for the first upload.

Step 4: Sort by date and follow the trail

You're looking for the earliest appearance. When you find matches, open them and note the dates. Then search the platform where the oldest copy appears, using words from its caption, to find the uploader. The original is usually longer, sharper and has more context than the viral copy.

Step 5: Search the words, too

Add text searches alongside image searches: a few distinctive words from the caption, the claimed place and date, plus terms like "video" or "footage". On YouTube and X, try searching with date filters set to before the clip went viral.

When you find nothing

No matches doesn't mean the video is genuine. It might be:

• New footage that hasn't been indexed yet.
• From a private or closed group.
• Generated by AI, so there is no earlier original anywhere.

That last case is exactly where a detector helps. Paste the link or upload the file to Video Verifier for a free Quick Scan. It checks the frames, the motion and the voice for AI traces, and reads any Content Credentials in the file.

Tip: if you find the original upload, scan that instead of the viral copy. Every re-upload loses detail, and the original gives the checks more to work with.

Put it together

Reverse search answers "has this been seen before, and where?". A detector answers "does this look made or altered by AI?". Used together, along with the 5-minute checklist, they catch the large majority of misleading clips that reach your phone.