Hey everyone, I'm Nicky, CEO of ScreenStop.
We're building Screen DLP — a new security category that traditional
DLP completely misses.
records, trading positions, or source code and photographs it with
their phone. The data never moves digitally — so network monitoring,
clipboard controls, and file restrictions are all blind to it.
What we built: on-device AI running on the endpoint's webcam that
detects when a phone is in capture position — and blurs the screen
before the photo is taken. No cloud, no data leaves the machine.
The hard engineering problem was false positive rate. Hands, coffee
cups, second monitors — all trigger naive models. Most of our work
went into making it not annoying to use.
Pre-seed, currently in pilots in healthcare and financial services.
Happy to discuss the detection model, the architecture, or the
category itself. What's your take — is this a gap you've seen in
your own security stack?
Tags: security, ai, machinelearning, privacy, DLP
Top comments (0)