Modelwire
Subscribe

Willison releases client-side face blur tool using MediaPipe

Simon Willison built a browser-based privacy tool that uses MediaPipe's computer vision models to detect and blur faces in photographs before sharing, addressing a real tension between documentation and consent. The tool runs entirely client-side via WebAssembly, meaning no images leave the user's device. This represents a practical application of edge ML for privacy protection, showing how vision models can be deployed for individual agency rather than surveillance. The work highlights growing demand for privacy-preserving alternatives to cloud-based image processing, particularly as photographers and activists seek to protect subjects' identities.

Modelwire context

Analyst take

Photo Scrubber is client-side, which means it removes the infrastructure dependency that makes surveillance and data harvesting economically viable. The real novelty isn't the blur capability itself, but that Willison distributed it as a tool for individual agency rather than as a backend service that could be repurposed for identification or tracking.

This directly counters the momentum in 404 Media's 'Surveillance Finds a Way' coverage from today, where Flock Safety's facial recognition push signals how computer vision vendors normalize intrusive capabilities as inevitable. Photo Scrubber operates on the opposite assumption: that individuals can retain control over their own visual data through edge deployment. The tension matters because it exposes a choice point the industry has largely obscured. Meanwhile, Meta's camera-free glasses from last week and Apple's on-device home security features represent similar bets on edge processing as a privacy hedge, but Photo Scrubber is the first tool we've covered that explicitly positions local processing as a defense against the identification infrastructure Flock and others are embedding.

If Photo Scrubber adoption reaches 100k monthly users within six months and law enforcement agencies begin requesting it as a standard evidence-handling tool, that signals a meaningful shift in how institutions think about visual consent. If adoption stalls below 10k, it suggests the friction of manual scrubbing outweighs privacy concern for most users, and the surveillance vendors win by default.

Coverage we drew on

This analysis is generated by Modelwire’s editorial layer from our archive and the summary above. It is not a substitute for the original reporting. How we write it.

MentionsSimon Willison · Photo Scrubber · Google MediaPipe · GPT-6 Astra · WebAssembly

MW

Modelwire Editorial

This synthesis and analysis was prepared by the Modelwire editorial team. We use advanced language models to read, ground, and connect the day’s most significant AI developments, providing original strategic context that helps practitioners and leaders stay ahead of the frontier.

Modelwire summarizes, we don’t republish. Simon Willison originally reported this story as “Photo Scrubber , local face blur & metadata removal”. The full content lives on simonwillison.net. If you’re a publisher and want a different summarization policy for your work, see our takedown page.

Related

Flock Safety pushes facial recognition into police camera networks

404 Media·

Facial recognition moves from infrastructure to real-time personal identification

404 Media·

Mayer's Dazzle bets photos over email for personal AI memory

Willison releases client-side face blur tool using MediaPipe · Modelwire