How to trace a photo’s origin with reverse image search (Google Lens and more). See if a profile photo is stolen or reused.
Free with no sign-up, and the analysis runs inside your browser — files never leave your device. On our public gate (30 real + 83 AI images) it measured 94.7% at original quality and 87.6% on recompressed copies, with zero real photos confidently mislabelled as AI. Treat it as an indicator, not proof: the newest commercial generators evade every detector.
A stolen photo almost always has its original somewhere on the web. Reverse image search is the strongest free tool for "where did this photo come from". It works on dating profiles, marketplace listings and circulating screenshots alike. This page shows the fastest route per device.
Android: long-press any image in Chrome and tap "Search with Google Lens". For gallery photos, use the Lens icon in the Google app. iPhone: open the photo, tap the info (ⓘ) button and use "Look Up", or search via Lens in the Google app.
Desktop: use the buttons below to open the Google Lens, TinEye and Bing upload pages, then drop your file in. Tip: if one engine finds nothing, try another — their indexes differ, and some matches only show up on TinEye.
Reading results: the same photo under a different name or country means theft. A stock-photo site means it is a purchasable photo, not "the profile owner". And finding nothing proves nothing — AI-generated photos have no original on the web. In that case, move on to the AI check on the home page.
A reverse image search is not one search. Engines index differently and their strengths diverge, so the same photo produces very different results. Google Lens is strong on objects, places and products; Bing is better at similar faces and compositions.
For tracing an origin, exact-match engines earn their place. They find pages carrying the same image, and where results can be sorted oldest-first you get far closer to where a photo first appeared. That sorting is exactly what a dispute over misuse needs.
A workable order: exact match first to find the earliest appearance, similar-image search next for variants, then object and place recognition to chase clues inside the photo. All three coming back empty is itself information for the next step.
Small changes defeat these searches: cropping, mirroring, filters, re-compression, an added watermark. To your eye it is the same photo; to the engine it is a different image. Anything that has passed through social media is already re-compressed.
There are ways around it. Crop out the background and leave the person or object large, then search again. Flip the image horizontally and retry — mirroring is a common way to evade search. Searching on one distinctive fragment often works well too.
Be careful how you read an empty result. Photos that appear nowhere fall into two groups: a genuine picture taken yesterday, and a generated image with no original anywhere. Reverse search cannot separate those, so at that point move the question to an AI-probability check.
Similar-image hits are not the same photo. Engines return matching composition and colour, so a different person with a similar face turns up regularly. Zoom in and confirm it is the identical image before claiming misuse.
A stock photo hit does not settle it either. People who modelled for stock photography do use their own pictures, and impostors do steal stock photography. Look less at the result itself and more at whether one image appears under several different names.
Read dates carefully. A result date can be a crawl date or the day a page was updated rather than the original publication. Use oldest-first to orient yourself, then open the page directly before treating a date as decisive.
A photo sent by a match returned catalogue shots from an overseas retailer and a stock-photo page, both years old. That single result ends the conversation. When one picture is scattered across commercial sites, the person sending it is almost certainly not its subject.
The listing image matched the manufacturer’s official product page exactly — not a photo of the seller’s actual item. That alone does not prove fraud, but it is reason enough to ask for real photos. An honest seller sends them within minutes.
AI-generated images have no original anywhere, so reverse search comes back perfectly clean. A genuine photo taken yesterday comes back exactly as clean. "No results" therefore appears in two opposite situations. When search is empty, move the question to an AI-probability check — reverse search can only answer "is this stolen?".
Search engines only accept images the user uploads directly. However, if you consent to research contribution during a home-page check, the reverse-search buttons on the result screen pass the stored original URL automatically — one click.
Uploaded images are handled under each engine’s policy. Think twice before uploading sensitive photos — every other check on this site runs entirely on your device.
Policies differ, and uploads may be used to improve the service. Think twice before submitting sensitive photographs. Note that the other checks on this site run on your device, while reverse image search is guidance for using external engines: a different arrangement.
It often works better. Removing the background and leaving the person or object large catches copies altered by cropping or filtering. Searching a single distinctive detail is worth trying too.