INITIALISATION DU SYSTÈME...
What the EXIF block inside an image holds, what GPS coordinates reveal, and how to read then remove that metadata without the photo ever leaving your browser.
EXIF data is a block of information your device writes inside the image file, with nothing showing on screen: the exact date and time, the camera model, the capture settings and, on a smartphone, the GPS coordinates of the location to within a few metres. It survives copying, emailing and cropping.
This article sets out what that block actually contains, what can be inferred from it, and how to read it and then remove it without the photo ever leaving your browser.
The Exchangeable Image File Format was established in 1995 to attach the technical details of a shot to the image itself. Designed for photographers, it became a privacy problem the day the camera turned into a phone with a GPS receiver.
A smartphone photo routinely carries several dozen tags. The most revealing are not the exposure settings but the identification and location fields.
| Field | What it holds | What it reveals |
|---|---|---|
| GPSLatitude / GPSLongitude | Coordinates of where the shot was taken | Home address, workplace, the children's school |
| GPSAltitude | Altitude at the moment of capture | The floor within a building |
| DateTimeOriginal | Exact date and time of capture | Routines, schedules, time away from home |
| Make / Model | Device make and model | The value of the equipment owned |
| BodySerialNumber | Camera body serial number | Links together photos published under different identities |
| Software | Application that saved the file | Shows that an edit took place, and with what |
A single coordinate points to a place. A set of photos points to a way of life. Cross-referencing the GPS fields and timestamps of images published over months reconstructs regular journeys, separates home from workplace by the time of day, and identifies the periods when a home stands empty.
GPSAltitude adds the vertical dimension: inside a building it narrows the search to a single floor. None of these inferences requires sophisticated tooling; they read straight out of the tags, and scripts extract them in bulk across thousands of files.
Stripping EXIF is not enough: an image file can carry several metadata blocks of different origins, which sometimes copy one another.
| Family | Who writes it | Typical content |
|---|---|---|
| EXIF | The camera or phone | Date, GPS, model, optical settings |
| IPTC | The photographer or newsroom | Author, caption, credit, usage terms |
| XMP | Adobe and similar software | Copies EXIF and IPTC, adds editing history |
| C2PA | Generative AI tools and some cameras | Signed provenance manifest: the tool that produced the image, declared edits |
A partial cleanup therefore often leaves the date and the author sitting in XMP while EXIF has been emptied. That is why GhostMeta works by re-encoding rather than deleting field by field: the image is redrawn from its pixels alone, and no block survives that operation.
The risk is not uniform. It depends on who receives the file and what they do with it. Three cases account for most of the exposure.
This is the most exposing case, because the photo is taken at home and shown to strangers. The GPS coordinates point at the address, and the timestamp shows when someone was there. A malicious buyer has an address and an inventory of what you own before sending a single message.
Unlike public platforms, an attachment transmits the file as is. No re-encode takes place: the recipient gets the full EXIF, GPS and body serial number included. This is the channel where cleaning beforehand matters most, and the one people think about least.
An image published once then circulates without you. If it carries a serial number, it links together every shot taken with the same device, including those posted under another handle years later. The quietest field is the most durable one here.
Cleaning the original file does nothing to copies already in circulation. The right sequence is the reverse of the usual one: take the post down first, then republish a cleaned version, rather than editing in place.
Before cleaning anything, look at what your files actually hold. It is the only way to measure your exposure rather than assume it, and to verify afterwards that the cleanup did happen.
The whole process runs in your browser: the image is decoded in memory, redrawn through the Canvas API, then exported again. No network request carries the photo, and the result holds no metadata block at all, by construction rather than by filtering.
For photos destined for an online listing, the listing photo security guide covers the procedure step by step. The security page explains how the local processing works.
The questions below cover EXIF and GPS. For image provenance, Content Credentials (C2PA) and how to verify a file after cleaning, see the questions hub.
No. Cropping changes the pixels, not the metadata block. The capture date, the device model and the GPS coordinates stay intact in the cropped file. Only a full re-encode of the image, or an explicit removal of the tags, strips them.
In a dedicated sub-directory of the EXIF block, the GPS IFD. Latitude and longitude are written there in degrees, minutes and seconds, each paired with a reference field (N/S, E/W) giving the hemisphere. Altitude, a GPS timestamp and sometimes the heading are often added. These fields are independent of the capture date: clearing one does not clear the other.
Most large platforms re-encode images on upload, which drops most EXIF tags from the published copy. That cleanup happens after the intact file, GPS included, has reached their servers. Stripping it before you upload remains the only way to control what leaves your device.
Removing the tags themselves does not touch the pixels. Any loss comes from the re-encode: a re-saved JPEG goes through one more compression pass. A good-quality re-encode makes the difference invisible to the eye, and lossless formats suffer none at all.
EXIF is written by the camera and describes the capture. IPTC is an editorial standard filled in by hand: author, caption, credit, usage terms. XMP is an Adobe container that can copy both and add the editing history. All three live in the same file and must be removed together.
No. Even with no coordinates, the device model, the software version and above all the camera body serial number remain identifiers. They are enough to link together every photo taken with the same device, including ones posted under different handles. Anonymity means removing the whole block, not just the location.