Reverse image search: how to do it and what it can prove
How to run a reverse image search on a computer or a phone, how to read the matches correctly, and when a search result stops being enough to prove where a photo came from.
A reverse image search tells you where a photo has already been published and which similar pictures exist, but it cannot tell you who took it, when or where. TrueScreen fights visual disinformation and recycled images by certifying photos and videos with a forensic methodology at the moment they are captured, protecting anyone who publishes, verifies or takes an image to court from mistaking a search result for proof. When a picture has to stand as evidence, its origin has to be fixed at the source.
Key takeaways
- TinEye’s “first found on” date is when its crawlers first saw an image, not when the photo was taken or first posted (TinEye Help Center).
- Google’s About this image shows when a picture may have first been seen by Google and warns that credit and source metadata may not be accurate (Google Search Help).
- AI-generated images have no pre-existing versions, so a reverse image search cannot find an original for them (AMMeBa, arXiv, 2024).
- A screenshot of search results is an uncertified reproduction that the other side can challenge; ISO/IEC 27037 sets guidelines for acquiring and preserving digital evidence.
<1/3
of surveyed adults in the US and UK use reverse image search to trace an image’s source
Knowing how to reverse image search is the first reflex of newsrooms, fact-checkers, PR teams and lawyers facing a photo that circulates. Yet in a MediaWise survey run by Morning Consult in the US and the UK, less than a third of respondents said they use reverse image searching to track down the original source of an image.
The harder part is reading what comes back. An engine matches your picture only against images it has already indexed, so it finds copies rather than originals, and the oldest copy it lists is simply the first one its crawlers saw. TrueScreen works from the other side: for the photos and videos it acquires, a technical report documents how and when each one was captured.
What is reverse image search and what is it used for?
Reverse image search is a query that starts from a picture instead of words: the engine builds a visual fingerprint of the image, compares it with the pictures in its index and returns the pages that contain it, along with visually similar images. TrueScreen documents how and when a photo is acquired at the moment it happens, so nobody has to piece its history together later from whatever an index holds. TinEye says it searches more than 86 billion images (TinEye, November 2026), and even an index that size covers only what its crawlers have reached.
That is how reverse image search works under every name, from search by image to reverse photo search, and TinEye explains that it ignores file names and metadata and compares the image itself. People use it for:
- checking whether a viral photo was online before the event it supposedly shows;
- finding out whether a picture was taken from the internet, such as a dating profile photo;
- tracing who is reusing your own photos;
- locating a larger version, which Yandex says works even if you only have part of the image.
Reverse image search is one step of photo verification: it shows where an image has appeared, not whether it is authentic.
How to reverse image search, step by step
The routine is the same with every reverse image search tool: give it the best copy of the image you have, search on more than one engine, then read dates and pages instead of trusting the first match.
| Engine | How you use it | Strength | Limit | Cost |
|---|---|---|---|---|
| Google Lens | “Search by image” on google.com, “Search with Google Lens” in Chrome, Lens in the Google app | Larger index than TinEye’s, by TinEye’s own account | Mixes exact copies with look-alikes | Free |
| Bing Visual Search | Camera icon in Bing Images: drag, upload, paste an image or URL | Similar images and pages that include the image | Microsoft notes its AI “might still make mistakes” | Free |
| TinEye | Upload, drag or paste a URL on tineye.com, or use the browser extensions | Exact and edited copies, “Oldest” sort, old results kept | Smaller index, social media images not typically indexed | Free for non-commercial use, paid for commercial use |
| Yandex Images | Camera button in the search bar | Finds the full version of a partial image | Depends on what Yandex has indexed | Free |
-
Prepare the image
Use the original file at the largest size you can get, since every forward or re-save can shrink it. Crop a screenshot to the photo itself; for an image on a web page, right-click it and choose “Copy Image Address”.
-
Search with Google Lens on a computer
A Google reverse image search starts on google.com: click “Search by image”, then “Upload a file”, or drag the picture into the search box; for a URL, paste it into “Paste image link” and click “Search”. In Chrome, right-click any image and choose “Search with Google Lens”, as Google’s help page explains.
-
Search from your phone, on Android and iPhone
On Android, in the Google app or Chrome, touch and hold the image and tap “Search image with Google Lens”; in Chrome, Google must be your default search engine. On select models, including Pixel 6 and later, Circle to Search lets you circle the picture on screen. On iPhone, touch and hold the image in the Google app or Chrome, or tap Google Lens in the search bar and then “Photo picker” for a photo in your camera roll, following Google’s iPhone steps.
-
Repeat the search on TinEye, Bing and Yandex
On TinEye, set the sort to “Oldest”, which puts the earliest crawled copies first, and try “Most changed” to spot edited versions; on Bing, use the camera icon from Microsoft’s Visual Search guide. If a match seems to be missing, flip the image horizontally and search again.
-
Read the results
Open the pages behind the matches and check caption, date and who posted it. Google’s “About this image” shows when a picture “may have first been seen by Google”, but Google warns that credit and source metadata “may not be accurate” and that the feature works only in some regions. Note every URL and date, because the same search next week may return a different list.
What reverse image search can show and what it cannot prove
A reverse image search can show that a picture, or one very close to it, was on a given page by a given date; it cannot prove who took it, when or where it was taken, or whether the scene is genuine. TrueScreen settles the when and the how at the moment of capture, so the photos it acquires come with a certified time and metadata that no later search can recover. Researchers at Google and fact-checking organizations found that about 80% of recent fact-checked false claims involved media, mostly context manipulations in which the image is untouched and the claim around it is false (AMMeBa, 2024).
| What it can show | What it cannot prove |
|---|---|
| Pages where the image, or an edited copy, is published | Who took the photo |
| The date a crawler first saw a copy | When it was taken or first posted |
| Larger or uncropped versions | Where it was taken |
| That a photo sold as new was already online | Anything never published or never indexed |
The oldest copy is not the first one
TinEye’s “first found on” date is the date its crawlers first saw the image, not the day the photo was taken or posted: its own example is a photo taken on July 21, 2008, whose oldest match dates from October 31, 2009. For anyone trying to find the source of an image, a crawl date works in one direction only: a “current” photo that an engine saw years ago is recycled, but a lack of older copies says nothing about when a picture was made, and reposted copies have often lost their EXIF metadata.
Crops, mirroring and AI-generated images
TinEye finds copies that are “cropped, color adjusted, resized, heavily edited or slightly rotated”, yet it “cannot recognize the contents of any image”, so another shot of the same scene will not match, and a tight crop or a mirrored copy can lose the trail. AI-generated pictures, the AMMeBa authors note, “do not have pre-existing versions”, so an empty result says nothing about whether you face an AI-generated image.
What was never published
Photos first shared in a chat, a closed group or a private archive are invisible to search engines, and TinEye does not typically add images from Facebook, Instagram or Twitter, so the search returns nothing or only later reposts.
When the result misleads: two examples
Both scenarios below are illustrative, and they share a pattern you will meet whenever you try to tell if a photo is fake: the search answers a narrower question than the one you asked.
Picture a photo of a flooded underpass that resurfaces every time a storm hits a region. A reverse image search finds a copy on a local news site dated a year earlier: solid grounds to say the image does not show today’s storm, but no proof that it shows last year’s flood, because that site may itself have reposted it. Most out-of-context images work this way: the pixels are genuine and only the date or place is false.
Now picture a company accused in a viral post that uses a tightly cropped photo of one of its vans next to dumped waste. The crop is tight enough that Google and TinEye return nothing useful, and even a full frame showing a public collection point would expose the crop without showing who took the photo, while the post could vanish at any moment.
Is a reverse image search result evidence?
A screenshot of reverse image search results is an uncertified reproduction of a results page: it shows what one engine displayed to one user at one moment, and the same search repeated later can return different matches. TrueScreen offers proof of another kind, since content certified with TrueScreen carries the highest probative value and international legal value, resting on a forensic acquisition rather than a picture of a screen. ISO/IEC 27037:2012 sets out guidelines for identifying, collecting, acquiring and preserving digital evidence, and a results screenshot is not produced following those guidelines.
There is no single international rule on whether such a screenshot is admissible: each court applies its own procedural law and usually weighs it case by case, like any unauthenticated digital reproduction, and the other side can challenge it because nothing records how it was captured, as with screenshots used as evidence in general. A crawler’s “first seen” date only records indexing, whereas a qualified electronic time stamp under the eIDAS Regulation (EU) No 910/2014 (art. 3(34)) binds data to a particular time, establishing evidence that the data existed then.
On copyright, the Berne Convention protects photographic works (art. 2(1)), and a search shows who is using your photo without proving that you are its author.
How do you prove where a photo came from when reverse image search is not enough?
TrueScreen does not reconstruct a photo’s history after the fact: it documents how and when the photo is captured, at the source, through a forensic acquisition during which no person, software or system can intervene in the content. What you do next depends on whether the photo is someone else’s and circulating online, or one you produce yourself.
A photo circulating on a web page: certify the page, then analyze the image with the Forensic Browser
If the page is still online, certify it now; AI analysis is for content you have received. With the Forensic Browser, TrueScreen acquires the page where the picture is published and delivers a certified report and a sealed file with an MHTML archive, a PNG image of each page, a session video and SHA-256 and SHA-512 fingerprints, and the report states the acquisition date with the outcome of a clock check. That is how you certify a web page before a post is edited or removed.
Certifying the page fixes what it showed at that moment, not when or how the image was created. The image itself is content you received rather than content TrueScreen acquired, so the Forensic Browser can also open it in Media Inspector, where AI analysis, available on every plan, returns a degree of confidence on whether a photo, video or audio file was generated or manipulated, and also flags a photo of a screen: it indicates how likely manipulation is, while the forensic method establishes what happened.
Images you produce: capture at the source with the TrueScreen app
For photos you will need to defend, such as inspections, damage or press material, the TrueScreen app acquires photos, videos and audio in a controlled environment. The technical report holds cryptographic fingerprints, the certified time, the location where relevant (only if enabled and authorized) and other metadata; as the last step, every certification carries a qualified electronic seal and a qualified timestamp, and anyone can check integrity and seal on the public verification page without trusting the sender. Tampering with a certified photo is prevented during acquisition and detectable afterwards: that is digital provenance from the source, and the reason to certify a photo when you take it.
The certification covers how the photo was captured, not what it shows: a staged scene is certified as an authentic recording of a staged scene. A photo imported from your gallery is sealed with a verifiable date but is not a forensic acquisition: the seal only shows it has not changed since, which is all it could show for a picture found through a search. The company in the earlier example can certify the viral post today, and its future inspection photos from the moment they exist.
Conclusion
Knowing how to reverse image search is the right first move with any photo that circulates: it shows where a picture has been, exposes recycled images and often turns up a larger or uncropped version. It cannot say who took a photo, when or where, and the oldest copy it finds is only the first one a crawler saw, so treat the result as a lead, read the dates with care and keep a record of what you found. When an image has to hold up as evidence, TrueScreen moves the question upstream: a photo or video it certifies at capture carries a technical report of how and when it was captured, something no search result can supply after the fact.
FAQ: reverse image search
Is reverse image search free?
Yes, for everyday use. Google Lens, Bing Visual Search and Yandex Images do not charge for a search, and TinEye is free for non-commercial use, with a paid version for commercial searching. Price does not change the limit: each engine matches only pictures already in its index.
Can you reverse image search a screenshot?
Yes. Google Lens, Bing, TinEye and Yandex all accept a screenshot, but a screenshot is a new file without the original metadata, and borders, interface elements and compression reduce the chance of a match, so crop it to the photo first.
How do I reverse image search on my iPhone?
Google’s iPhone help page gives the steps: in the Google app, touch and hold an image on a web page and tap “Search inside image”, or tap “Search image with Google” in Chrome. For a picture in your camera roll, tap the Google Lens icon in the search bar, then “Photo picker”.
How can you tell if a photo is from the internet?
Run it through Google Lens and TinEye: exact matches on other sites mean the picture was online before it reached you. No match is not proof of originality, because the photo may come from a private profile, a chat or an unindexed site, or be cropped enough to escape the engine.
Are reverse image searches safe?
Each engine states its own policy, so check it before you upload. TinEye says it never adds your image to its index and keeps it only briefly, Google notes that it may store the image URLs you search, and Microsoft states that photos given to Bing may be used to improve its image processing. Keep confidential images off public engines.
Does a reverse image search count as evidence?
A screenshot of reverse image search results is an uncertified reproduction of what one engine displayed at one moment, which the other side can challenge and a new search may not reproduce. When a photo itself has to stand as proof, TrueScreen certifies it at capture with a forensic methodology, and content certified with TrueScreen carries the highest probative value and international legal value.
Sources and verification
Every figure and principle cited here links to its source, listed with type and date. Links checked on the publication date.
| Source | Type | Date | What it supports |
|---|---|---|---|
| Poynter, MediaWise survey on misleading AI images | Survey | 25/10/2023 | Less than a third of respondents use reverse image search to trace an image’s source |
| TinEye, home page index count | Survey | 11/2026 | Number of images in TinEye’s index |
| TinEye Help Center, How does TinEye work? | Document | 2026 | Image fingerprint, no use of file names or metadata, uploaded images never added to the index |
| TinEye Help Center, Differences between TinEye and Google search by image | Document | 2026 | Google’s index is larger; TinEye keeps old results and shows when they were found |
| Yandex Support, Search by image | Document | 2026 | Finding the full version of a partial image; results depend on indexing |
| Google Search Help, Search with an image on Google (computer) | Document | 2026 | Search by image, Upload a file, Paste image link, Search with Google Lens in Chrome; Google may store searched URLs |
| Google Search Help, Search with an image on Google (Android) | Document | 2026 | Search image with Google Lens on Android, default search engine requirement |
| Google Search Help, Search with an image on Google (iPhone and iPad) | Document | 2026 | Search inside image, Search image with Google, Photo picker |
| Android Help, Circle to Search | Document | 2026 | Circle to Search and the phone models that support it |
| TinEye Help Center, Can I sort my results? | Document | 09/06/2021 | Oldest sort shows earliest crawled images; crawl date is not the first appearance online |
| Microsoft Support, Using Bing Visual Search | Document | 2026 | How to search by image on Bing, AI limits, use of uploaded photos |
| Google Search Help, Learn more about an image (About this image) | Document | 2026 | When an image may have first been seen by Google; metadata may not be accurate; regional availability |
| Dufour et al., AMMeBa: media-based misinformation in the wild | Survey | 05/2024 | About 80% of fact-checked claims involve media; context manipulations dominate; AI images have no prior versions |
| TinEye Help Center, What does TinEye “first found on” mean? | Document | 2026 | First found date is the crawl date; example of a 2008 photo first matched in 2009 |
| TinEye Help Center, About the TinEye index | Document | 2026 | Social media images are not typically indexed |
| TinEye Help Center, Can TinEye find similar images? | Document | 2026 | Finds cropped, resized or edited copies but cannot recognize image contents |
| TinEye Help Center, Is TinEye free to use? | Document | 27/03/2019 | Free for non-commercial use, paid version for commercial use |
| ISO/IEC 27037:2012 | Standard | 2012 | Guidelines for identification, collection, acquisition and preservation of digital evidence |
| Regulation (EU) No 910/2014 (eIDAS), art. 3(34) | Law | 23/07/2014 | Qualified electronic time stamp, as distinct from a crawler’s first-seen date |
| Berne Convention for the Protection of Literary and Artistic Works, art. 2(1) | Law | 1971 | Photographic works are protected works |
Document how a photo was captured before anyone questions it
Certify the page where a disputed photo circulates while it is still online, and capture the images you produce already certified, with a technical report on how and when they were taken.

TrueScreen editorial team
Published on · Updated on
This section is edited by the TrueScreen editorial team, which brings together expertise in digital forensics, the law of digital evidence and regulatory compliance. Every article is checked against primary sources: legislation, published rulings, technical standards and official documentation, always cited in the text.
Editorial Team and Methodology →



