Guide Β· Media Literacy
The fastest way to tell whether a viral photo is what it claims to be is to find where it came from. Reverse image search does exactly that β and once you know which engine to reach for, it takes about thirty seconds.
Recycled photos are the workhorse of online misinformation. An old fire, a protest from a different country, a movie still passed off as breaking news β these spread not because they're sophisticated, but because almost nobody checks. Reverse image search is the check. You hand an engine a picture instead of words, and it tells you where else that picture lives on the web.
Investigators at Bellingcat treat it as a baseline step before an image is trusted or published. Here's how to do it properly β including the parts most quick tutorials skip.
No single engine finds everything, and their indexes barely overlap. If you only ever use one, you're missing whatever the others catch. Match the tool to what you're trying to learn.
Built into Chrome, Android, and the Google app, so it is the fastest to reach. Good at surfacing news articles and stock-photo listings that reuse the same shot. It deliberately blocks matching a person's face across photos, so don't rely on it for identifying people.
Increasingly the fact-checker's first stop for tracing a photo back to its source. Its filters β by size, layout, date, and license β help you jump straight to the largest (usually the earliest) copy rather than a cropped repost.
The strongest free engine for matching faces and near-duplicates, and it routinely finds copies on Eastern-European sites that Google and Bing miss. When the other engines return nothing, this is the one to try next.
The original reverse-search engine (2008) with an index of tens of billions of images. It excels at exact and near-exact matches and lets you sort results by oldest β the single most useful button when you're checking whether a photo is being recycled.
On a computer: the quickest route in Chrome is to right-click any image and choose "Search image with Google." For the other engines, save the picture (or copy its URL) and drop it into the upload box on Bing, Yandex, or TinEye. All four accept a file, a pasted image, or an image URL.
On a phone: screenshot the image first. In the Google app, tap the Lens icon and pick the screenshot from your gallery. For Yandex or TinEye, open the site in your mobile browser and use their upload button β both work fine without an app. If a photo lives inside a chat app, save it to your camera roll before searching; sharing directly often passes a compressed version that matches worse.
This is the step that separates a shrug from a find. Instead of searching the whole busy scene, crop tight on one telling element β a street sign, a storefront, a logo on a uniform, a face, a distinctive building. Engines match a single clear subject far more reliably than a cluttered frame, and a readable sign or landmark can pin down a location the caption is lying about. Google Lens and Yandex both let you drag the crop handles right in the results view, so you can re-search a smaller region without starting over.
The earliest appearance of an image is usually its true origin. On TinEye, sort results by "Oldest." On Bing, filter by date. If the same photo turns up in a 2019 news story, a "breaking now" caption in 2026 is misleading β even if the photo itself is real. This time-stamp check is the fastest way to catch the most common trick in the book: a genuine old image wearing a fresh, false caption.
A match is a lead, not a verdict. Click through to the pages that host the image and read what they actually say. Does a reputable outlet caption it as a different event, a different year, a different country? Is the "original" itself just another unsourced repost? Trace the chain back until you reach a credible first publisher β a wire agency, a named photographer, an official account. If every copy dead-ends in anonymous accounts, treat the image as unverified rather than confirmed.
The image is a screenshot from a video
Reverse search rarely catches video frames directly. Grab a clean keyframe (pause and screenshot, or use InVID-WeVerify to extract frames), then search that. Try several frames β a face turned to camera works better than a blurry motion shot.
It's been cropped, mirrored, or filtered
Edits break exact-match engines. Flip the image horizontally and search again, or crop to an unedited corner. Yandex tolerates transformation better than most, so escalate to it when Google and TinEye come up empty.
Every result is from the last few hours
If the only hits are today's reposts, you likely can't establish age from the image alone. Widen the query with a distinctive detail β a banner, a jersey number, a shopfront name β and search those words alongside the picture.
Nothing matches at all
A genuinely new photo returns nothing β but so does an AI-generated one. If the picture has that too-smooth, plausible-but-placeless look, stop treating it as a photograph and switch to AI-image checks instead.
Face-matching engines are powerful, and that cuts both ways. Use them to check whether a "profile photo" was lifted from a stock library or someone else's account β a routine catfish and impersonation check. Don't use them to unmask or track private individuals. The goal of verification is to test a claim, not to surveil a person.
Already have a claim in mind, not just an image? FAXTR searches 100+ fact-checking organizations across 11 languages in one box β free, no login β so you can see whether a viral photo or story already has a published verdict.
Go to the verifier β