A photo of a flood circulates on WhatsApp as if it were taken yesterday. Is it? The fastest question to answer is not “is it fake?” but “where has it appeared before?” That is exactly what a reverse image search does: you upload the picture instead of typing keywords, and the engine returns pages where that image, or a very similar one, is published.
This matters for anyone who wants to check an image before sharing it. Google Lens surfaces websites containing the photo and its context; TinEye returns exact or edited copies sorted by age; Bing and Yandex serve as a second opinion. None of them tells you whether an image is true — they tell you where it has been. And that is often enough to dismantle a hoax.
- Context over authenticity: Reverse search exposes recycled real photos, the most common visual fraud, rather than AI-generated fakes.
- Different engines, different indexes: Google Lens, TinEye, Bing and Yandex crawl different corners of the web, so using at least two is recommended.
- Metadata matters: Google can display IPTC credits and flag C2PA credentials or SynthID watermarks when present.
- Oldest date is a clue, not proof: TinEye’s “oldest” result reflects when its crawler first found the image, not when it was originally published.
What a reverse image search actually tells you
It tells you where an image, or a close variant, is published — not whether what it depicts is accurate. If a photo presented as “today” appears in a news story from years ago, the hoax is exposed without analysing a single pixel.
This is the right tool for the most common fraud, which is not the AI-generated image but the real photo reused out of context: a fire from another country, a protest from another decade. Synthetic images require other signals, such as deepfake detection techniques for video, voice and stills.
The practical recommendation is to use at least two engines. They do not index the same material or search the same way, and what one misses the other often finds.
Google Lens: the most versatile, but built around “similar”
Google Lens returns objects appearing in the image, similar images and websites containing that image or a similar one. Google’s official help page on searching with an image adds that results may also include AI summaries.
How to use it, according to that same documentation:
- On desktop: on google.com, click the image search button and upload a file, drag it into the search box or paste the image URL. In Chrome, right-click any image and choose “Search with Google Lens”.
- On Android: from the Google app or Chrome, long-press an image on a web page and select “Search image with Google Lens”. You can drag the corners of the selection box to search only part of the image.
- On iPhone: from the Google app, Chrome or Safari, which also allow searching with a photo from your camera roll.
One privacy detail the help page does not hide: browsers do not store searched URLs in history, but Google may retain them to improve its products.
The most useful verification feature is not in the results list but in the image panel. According to Google’s page on image information in Google Images, it can display credits, copyright and licence data drawn from IPTC metadata, and indicate whether an image was generated with AI or captured with a camera when C2PA data or a SynthID watermark is present. Google began embedding those credentials in 2024.
TinEye: exact copies and an age clue
TinEye searches for copies of the same image, even if cropped, resized, recoloured or slightly rotated, and does not search for similar images. Its FAQ explains that it creates a digital fingerprint of the photo and compares it against its index, without using the file name or metadata. As of November 2026, its homepage cites an index of more than 86.6 billion images.
That limitation is its virtue: because it does not recognise content, it will not show you other fires — only that fire.
What makes it valuable for debunking:
- Results can be sorted by best match, largest image, most modified, newest or oldest.
- The “Compare” function toggles between your image and the result to reveal crops or edits.
- It accepts JPEG, PNG, GIF, BMP, TIFF and WebP up to 20 MB. It works best with images at least 300 pixels on a side and accepts from 100.
- It states that it does not store or share the images you search.
One caveat TinEye itself stresses: the “oldest” date is when its crawler found the image, not when it was first published. If you find the photo in 2019, you know it existed in 2019. You do not know whether it existed earlier.
Avoid images with visible watermarks, since the engine may end up searching for the watermark. Extensions are available for Firefox, Chrome, Edge and Opera.
Bing and Yandex as a second opinion
They fill gaps: each engine crawls different sites, and a photo missing from Google may appear in another index.
Bing Visual Search lets you drag an image, take a photo or paste an image or its URL, and promises similar images and object identification. It is integrated into the right-click menu in Edge, the Windows 11 Snipping Tool and Windows Photos, and has extensions for Chrome and Firefox. Microsoft warns that availability varies by device, market and browser version.
Yandex Images remains accessible at yandex.com/images. Its official help states that it can find copies at different sizes or similar images, with a “Similar” button in the results. As with any service you upload a photo to, check its privacy policy first if the image is personal.
A five-step method that almost always works
- Get the best version of the image. A screenshot cropped to the photo, without borders or added text, yields better results.
- Start with Google Lens to see the context: which sites carry it and what they say about it. Open the image panel in case content credentials are attached.
- Run it through TinEye and sort by “oldest”. If the photo existed long before the event it claims to show, you already have your answer.
- Cross-check with Bing or Yandex if the first two find nothing.
- Read the pages, not the thumbnails. A photo appearing on a news site does not mean the site uses it with the same caption as the message you received.
If you suspect the image is synthetic, Google offers a verification tool in the Gemini app that looks for SynthID marks and content credentials. It has a clear limit that Google acknowledges: it only recognises SynthID in content created with its own tools. Failure to detect it does not mean the image is real. And if the photo comes from a social network, also check whether it carries the “made with AI” label used by Instagram, Facebook, TikTok or YouTube — with the same caution: its absence proves nothing either.
Reverse image search does not deliver certainties. It delivers provenance. That is precisely what is missing from most visual hoaxes in circulation: nobody has asked where the photo came from. Thirty seconds with two search engines resolves more cases than any AI detector, because the cheapest lie remains a real photo attached to a false caption.