The photo of the bike. And your address.
This is not a threat model about state adversaries. It is a photo of a bicycle for sale, taken in your driveway, uploaded to a public listing, carrying the coordinates of your driveway to within a few metres.
Where this actually bites
The everyday cases are the ones that catch people, because there is no moment where it feels like a security decision:
- Selling something online. A photo taken at home, posted publicly, to an audience that by definition includes strangers who want to come and collect it.
- Posting to a forum or a hobby community. Months of photographs from the same place, adding up to a map.
- Sending a picture of a document. The coordinates come along, and so does the uncropped thumbnail if you cropped anything out.
- Sharing photos of children. The school gate, the park, the front door, each with a timestamp.
Some platforms strip metadata on upload. Some do not. Some strip it from the version they display and keep the original. You cannot tell which from the outside, and you cannot undo an upload that did not.
What ScrubPony changes
- One command over the folder you are about to upload.
scrubpony -r -d ~/clean ~/Pictures/to-uploadgives you a clean copy with the same names and structure. The originals are not touched. - It tells you what it found. The summary counts how many files carried GPS. That number is usually higher than people expect, and it is the whole reason the line prints even when nothing went wrong.
- Your portrait photos stay upright. A 36-byte orientation block is written back automatically, so you do not end up with a folder of sideways photographs — which is the reason most people abandon metadata strippers.
- Colours do not shift. The ICC profile is kept by default.
- The picture is untouched. No re-encoding, no quality loss, so the clean copy is as good as the original.
The quickest version
If you only ever run two commands, make them these:
scrubpony -n -r ~/Pictures/to-upload— look at what is in there. It writes nothing.scrubpony -r -d ~/clean ~/Pictures/to-upload— make a clean copy and upload from~/clean.
Also worth doing once, on your phone: turn off location for the camera, so the next few years of photographs do not need this at all.
Honest limits
It needs a Mac or a Linux computer and a command line. If your photographs live only on your phone, the built-in "remove location" option in the share sheet is the realistic answer — and here is exactly what it does and does not cover. A tool you will not use protects nothing.
It handles JPEG only, and most modern iPhone photos are HEIC. Exporting as JPEG first works; skipping the step means the file is reported and skipped rather than cleaned.
Most importantly: this does not make a photograph anonymous. It removes the labels, not what is in the picture. A house number, a street sign, a reflection in a window, a school uniform, or a distinctive view out of the back is all still there, and no tool can help with that. Look at the picture before you post it.
And it cannot help with what has already gone. Scrub before you share, not afterwards.
Common questions.
Do I really need this if the site strips metadata?
Some sites do, some do not, some strip it from the copy they display while keeping the original, and none of them tell you which. Stripping it before upload is the only version of this you control.
Is it hard to use?
Two commands, and the first one only looks. If a terminal is not something you use, your phone's "remove location" option is the honest recommendation instead.
Will my photos look different?
No. The image data is copied through byte for byte, the orientation is preserved, and the colour profile is kept by default. The file is smaller by exactly the size of what was removed.
Should I just turn off location on my camera?
Yes, and it is the better fix going forward. It does nothing for the photographs you already have, which is what the scrubber is for.
Get ScrubPony
Strips identifying metadata out of JPEGs without touching a pixel. C99, no dependencies beyond libc, macOS and Linux, MIT-licensed.