Photo Scrubber — local face blur & metadata removal
Simon Willison’s Weblog Simon Willison
Simon Willison built a local tool that blurs faces in photos. It also strips metadata, so strangers stay a lot less identifiable.
Based on reporting by Simon Willison’s Weblog, Simon Willison — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Simon Willison has put together an experimental tool called Photo Scrubber for one very specific problem: what happens when a photo includes strangers you don’t want to expose online. He says the trigger was a picture of protesters, and the discomfort of sharing images when people in them can still be identified by face.
The tool is designed to run locally. It finds faces automatically and blurs them before the image goes anywhere else. That matters because the privacy fix happens on the machine, not after the file has already been uploaded and copied around.
Under the hood, it uses Google’s MediaPipe C++ library, compiled to WebAssembly through @mediapipe/tasks-vision, along with the BlazeFace face detection model. Those are the moving parts doing the detection work; the result is a practical little pipeline rather than a grand platform pitch.
There’s also a metadata angle. Photo Scrubber removes metadata too, which is the kind of detail people forget until a photo quietly hands over more than the pixels they meant to share. The whole thing is still marked experimental, but the intent is clear: make sharing a little less careless without sending the image off to a cloud service first.
My take — AI-written commentary, not fact-checked reporting
Local privacy tools are the right kind of boring: useful, specific, and a lot less theatrical than another cloud promise. Face blur plus metadata stripping is the sort of feature every photo app should have by default, not as a bonus after the damage is done.
Read more about this at: Simon Willison’s Weblog