What a photo carries besides the photo
A picture taken on a phone ships with a block of data describing the circumstances it was taken in. Where you were, to about five decimal places. Which device, often down to a serial number. The exact second. Sometimes your name, if you ever filled in an owner field.
None of that is visible when you look at the image, which is the problem. It travels attached to the file, and it goes wherever the file goes.
Why this matters more than it sounds like it does
The classic version is someone selling furniture online, photographing it at home, and publishing the coordinates of their living room to a marketplace. It has caught out enough people to be a genre. The same applies to a photo of a pet in the garden, a work-from-home desk setup, or anything else shot somewhere you would rather not pin on a map.
The device serial is quieter and does something different: it links otherwise unconnected uploads to one camera. Anyone comparing two files can tell they came from the same device, which is exactly the property you do not want across accounts you keep separate.
| Field | What it reveals | Removed here |
|---|---|---|
| GPS location | Where the photo was taken, to about five decimal places | Yes |
| Device model | Which phone or camera, sometimes a serial number | Yes |
| Date and time | The exact second of capture | Yes |
| Owner and artist | A name, if the field was ever filled in | Yes |
| Colour profile | How colours should be displayed. Nothing about you | No, deliberately |
The thing most of these tools get wrong
Nearly every remove EXIF tool works the same way. It draws your image onto a canvas and exports it again. That does remove the metadata, because the export writes a completely new file, and it also re-encodes every pixel. You hand over a photo and get back a measurably worse one, usually with no mention that this happened.
A JPEG does not need that treatment. It is a sequence of marker segments, and metadata sits in its own segments, separate from the image data. Drop those segments, join up what is left, and you have the same image with the metadata gone. Not visually identical, actually identical: the compressed image data is copied through untouched.
This tool does that. PNG works the same way with different containers, so the text and Exif chunks come out and the rest stays.
The colour profile stays, on purpose
An ICC profile is technically metadata and gets swept up by tools that remove everything they can find. It describes how colours in the file should be interpreted, so removing it is what makes a cleaned photo come back looking washed out or oddly saturated on someone else's screen.
It also contains nothing about you: no location, no device identity, no timestamp. There is no privacy argument for removing it and a clear visual cost, so it stays.
What this does not do
It does not remove an AI watermark, and no browser tool does. Systems like SynthID mark the pixels rather than the metadata, and surviving re-encoding, cropping and compression is what they are specified to do. Stripping metadata leaves them entirely intact. If a tool implies otherwise, it is describing metadata removal in more exciting words. We wrote about what actually survives separately.
It also cannot un-publish anything. Metadata already uploaded somewhere has already been read. This is a step to take before a file leaves your machine, not a way to retract one that has.
Canvas re-encode (most tools)
- Draws your image and exports a new file
- Metadata is gone because the file is new
- Every pixel is recompressed
- The photo comes back measurably worse, usually unmentioned
Segment removal (this tool)
- Drops the metadata segments and rejoins the rest
- Metadata is gone because it was removed
- No pixel is touched
- The image data is byte for byte what you supplied
A note on what platforms already do
Most large social platforms strip metadata from uploads themselves, as a side effect of re-encoding everything for their own storage. So a photo posted to Instagram is generally not leaking coordinates to people who download it from there.
The gap is everywhere else: marketplace listings, forums, a file emailed to someone, a direct download from your own site, an image attached to a support ticket. Those routes commonly pass the original file straight through. If you cannot say for certain which kind you are dealing with, strip it first, because the check costs nothing and the failure is permanent.