Ent launches an on-device workspace security platform that uses local models to automatically obscure sensitive screen content

This runs locally and watches what's on your screen.

If you're about to expose something sensitive (like sharing your screen in a Zoom call with a private doc open), it automatically covers it before it's visible.

You can always one-click reveal it if you actually meant to share it, but the agent protects you by default.

This is really cool.

They say this is all based on what I'm trying to do (my "intent"), not just what's on my screen, but I honestly have no clue how they can determine that. I tried to find more details, but couldn't.

Regardless, if they figure that out, this is going to be huge for privacy and security. I can see every company wanting this on every employee's laptop.

@kimmonismus Or… you could use FreeLattice and own all your data?

Why let anyone sell your data? •

@kimmonismus that model changes the endpoint cat and mouse game finally.

does the local processing hit latency for real-time meetings?

@kimmonismus intent over keywords is a huge leap

@kimmonismus Shifting from post-breach scanning to pre-exit intent classification is the hardest pivot in endpoint security. The real metric will be false-positive rate on those local intent models vs. traditional DLP alert volume.