What Actually Lives Inside Your Photo's Metadata
Most advice about EXIF stops at "it might contain your location". That undersells it. A single JPEG straight off a phone can carry a few dozen tags describing the device, the moment, the settings and sometimes the photographer, and a few of them are far more identifying than a pair of coordinates. Here is the inventory, and a framework for deciding when to clear it. The step-by-step removal guide covers the mechanics; the Metadata tool lets you read the tags first.
The four groups of tags
Location
The GPS block is more detailed than most people realise. Beyond latitude and longitude it can record altitude, a UTC timestamp independent of your device clock, the direction the camera was pointing, your speed at the moment of capture, and an estimate of the positioning error. Direction plus coordinates is enough to work out which window a photo was taken from, not merely which building.
Device
Make, model, lens model and the software or firmware version are standard. Many cameras also write a body serial number and a lens serial number. Serial numbers are the quiet problem: they link every photo you have ever published to the same physical camera, which is how anonymous image sets get correlated with a named owner who once posted a warranty photo. Vendor-specific MakerNote blocks can add shutter count, focus points and occasionally an owner name typed into the camera menus years ago.
Time
There are usually several timestamps: when the shutter fired, when the file was digitised, when it was last modified, sub-second precision, and in newer files the UTC offset. Together they reveal your time zone and your daily routine across a photo set even if you never geotagged anything.
Attached extras
EXIF is not the only passenger. XMP packets from editing software record the edit history, the application used and sometimes the original file path, which can include your username and folder names. IPTC fields carry creator, credit and contact details for press work. And an embedded thumbnail is stored separately from the main image. There are documented cases of photographs published after cropping that still contained a thumbnail of the uncropped original, because the editor updated the image but not the preview.
Redacting pixels does not touch any of this. Blurring a face leaves the GPS coordinates, the camera serial and the timestamp fully intact, and can leave an embedded thumbnail showing the unblurred frame. Metadata stripping is a separate pass. The article on choosing a redaction method covers the pixel half of the job.
When stripping is the right call
- Marketplace and classified listings. Photos of furniture taken inside your home are the classic leak: the item sells, the coordinates stay.
- Anything involving children. School uniforms plus home coordinates plus a timestamp is a pattern you do not want assembled by a stranger.
- Dating profiles and forums. Both are places where you are talking to strangers who can download the original file.
- Direct file transfers. Large social platforms usually re-encode uploads and drop EXIF from the public copy, but email attachments, chat file sends and shared cloud links deliver your file byte for byte. Those are the risky channels, not the feed.
- Sensitive documents. Scans and screenshots of paperwork inherit the same tags, and if you have run them through OCR the image is probably about to be shared with someone.
When you should keep it
Blanket stripping is not always right. Photographers publishing to a portfolio lose the aperture, shutter and ISO readings that make their work useful to other photographers. Family archives lose the dates that make them sortable in twenty years. Insurance and legal evidence relies on capture timestamps, and a stripped file is a weaker exhibit than an intact one. Licensed press images need their IPTC creator and credit fields to survive. The useful habit is to keep an untouched master in your own storage and strip only the copy you publish.
Two side effects worth knowing
Orientation. The rotation of many phone photos is not baked into the pixels; it is a single EXIF flag that viewers read and apply. Remove all EXIF and a portrait photo can suddenly display on its side in software that was relying on that flag. If it happens, rotate the image properly with the rotate tool so the correction is in the pixel data itself.
Colour. Colour profiles are technically metadata too, but removing them can shift how an image renders. The cleaner is built to keep the chunks that describe colour and transparency while dropping the descriptive ones, so a cleaned PNG or WebP still looks the way it did.
Read before you erase
The most useful thing the tool does is not the removal, it is the table it shows you first. Load a few photos from your own library and look at what is actually there - people are consistently surprised. Since the parsing and the cleaning both happen inside your browser, you can audit genuinely sensitive files without handing them to a server just to find out what they were leaking.