Most phone photos carry far more data than the picture shows. Inside the file, alongside the pixels, sits a block of metadata: the GPS coordinates where the photo was taken, the phone or camera model, the exact date and time, the software that edited it, sometimes even the author's name and a thumbnail preview of the original. Anyone who saves the image can read all of it with free tools — and the photo itself gives no hint that any of it is there. The fix takes seconds: strip the metadata before the photo leaves your device. Our free Photo Privacy Cleaner does exactly that in your browser — drop in a photo, and it rebuilds the image from its pixels alone, then proves the metadata is gone with a before-and-after scan. Nothing is uploaded at any point. This guide explains how to remove photo metadata with it, why rebuilding beats editing, the holiday-photo example that shows what is really at stake, and what no metadata tool can remove.
Step one: drop in your photo. Drop a photo onto the page, or click to browse your device — JPG, PNG or WebP. Phone photos, camera exports, screenshots with data attached, anything in those formats. The file stays on your device from start to finish. Step two: the tool rebuilds it from pixels. Your browser decodes the photo into pixels, paints those pixels onto a fresh canvas, and encodes a brand-new file from the canvas. Metadata containers never make the trip — only the picture itself survives. Step three: verify, then download. The cleaned file is immediately re-scanned with the same scanner as our AI Label Checker, and you see both reports side by side — the original file's metadata next to the cleaned file's — before you download. If the scanner finds nothing in the cleaned file, you are not being asked to trust a promise; you are looking at the same evidence a platform or a stranger would find.
Most "metadata removers" edit the file in place, deleting tags one by one — and any tag the developer forgot to handle survives. This cleaner takes the stronger approach: it throws the original file's structure away entirely. A canvas has nowhere to put EXIF, XMP, IPTC or C2PA data, so none of it can leak into the output — GPS coordinates, camera and lens details, software names, timestamps, author fields, edit history and embedded preview thumbnails are all simply absent from the new file. That is also why the verification step matters. After rebuilding, the tool feeds the new file's bytes back into the same metadata scanner that powers the AI Label Checker — the scanner that looks for C2PA manifests, IPTC fields, EXIF GPS and the rest — and displays the two reports next to each other. ExifTool's own documentation warns that metadata is a maze of overlapping standards — MakerNotes, proprietary chunks, XMP sidecars — and cleaner apps routinely miss a container or two. The rebuild approach sidesteps the maze instead of navigating it: there is no list of tags to forget, because nothing is copied.
A parent photographs their child in the garden and posts it to a school parents' group. The photo looks harmless — but the file's EXIF block carries the phone model, the exact date and time, and GPS coordinates accurate to a few metres of the family home. Anyone who saves the image can read all of it with free tools. Run through this cleaner first, the before report lists the GPS block, the device fields and the timestamps; the after report is empty, and the picture itself is unchanged. Multiply that by the everyday cases — marketplace listings photographed at home, a "for sale" car shot on your own driveway, a workplace photo with a location-tagged whiteboard — and the habit of cleaning before sharing stops a whole category of accidental disclosure. The same thirty seconds protect a job listing photo taken at your desk, a product shot taken in your living room, and a school photo taken in your front garden: the picture says what you meant it to say, and nothing else.
A tool that says "cleaned" and shows you nothing is asking for trust. This one shows you the evidence: the before report from your original file, the after report from the rebuilt file, produced by the same scanner, side by side. If the after report is empty, the metadata is gone — not hidden, not promised gone, but absent from the actual bytes you are about to download. That distinction matters because metadata hides in containers most people have never heard of: a C2PA manifest that records the editing chain, an IPTC block with author and location fields, a MakerNote that only one camera brand writes, a preview thumbnail that is itself a small copy of the original image. An editor that deletes tags by name has to know every one of those containers to be safe. A rebuild does not have to know any of them, because it never copies the file at all — and the re-scan then checks the result instead of the intention.
One thing the cleaner deliberately does not touch: anything burned into the pixels themselves. A house number visible in the photo, a reflection in a window, a street sign, or an invisible AI watermark such as SynthID all live in the image, not the metadata, and no metadata tool can remove them. For visible sensitive details, cover them first with the Screenshot Redactor or Face Blur, then clean the result here — redaction first, metadata second, in that order. And the cleaner covers photos only. Working with a video instead? Run the video through the AI Label Checker to at least see exactly what metadata it carries before you post it. The same habit applies: look at what a file carries before you share it, clean what can be cleaned, and cover what cannot.
Trusting an in-place editor. A remover that deletes tags one by one can miss a container — MakerNotes, proprietary chunks, XMP sidecars — and the missed container still carries your location. Rebuild, then verify. Sharing the original alongside the cleaned copy. The cleaned file is safe; the original attached to the same message, or kept in the same shared album, is not. Upload only the cleaned export. Forgetting screenshots carry data too. A screenshot of a map, a photo app, or a camera roll can carry its own metadata and visible location details — clean the screenshot the same way, and redact visible details first. Assuming the platform strips everything. Social platforms often strip some metadata on upload, but not always all of it, and anyone who downloads the file before it is uploaded — or receives it by message, email or a shared drive — sees everything. Clean before you share, not after you post. Cleaning after posting. Once a photo is public, copies already exist; cleaning the file on your phone does nothing for the copies already downloaded. The thirty seconds have to happen before the upload.
The honest limits are three. First, the cleaner works on photos — JPG, PNG and WebP — not on video; for video, the AI Label Checker shows you what metadata it carries, which is the first step before deciding how to share it. Second, the work happens in your browser's canvas engine, so extremely large images on an old phone can be slow, though normal phone and camera photos clean in a second or two. Third, the rebuild removes metadata, not pixels — anything visible in the picture itself stays visible, and anything invisible-but-in-the-pixels, like a SynthID watermark, stays too. Within those limits, the tool does one job completely: the file you download carries the picture and nothing else, and the before-and-after scan proves it. Two companion guides cover the neighbouring jobs: our AI label checker guide explains how to read the markers — C2PA, IPTC and EXIF — in any photo or video, and our screenshot redaction guide covers blacking out names, balances and keys before you share.
Check a photo or video for AI markers — C2PA Content Credentials, IPTC and EXIF.
Paste a screenshot, drag boxes over names, balances and keys, and export a clean PNG.
Auto-detect faces or draw your own boxes before you share.