Everyone wants AI to understand the real world But nobody wants their face randomly ending up in some AI training dataset That's one detail I found interesting while digging into @vangrid_io According to their docs, faces and license plates are blurred directly on the contributor's phone before footage gets uploaded And buyers receive the reconstructed 3D model instead of the original video Sounds like a small detail, but it's actually pretty important Physical AI needs to understand streets, buildings and obstacles, not collect everyone's personal information Of course, automatic blurring isn't perfect But handling privacy before data leaves the device makes a lot more sense than trying to fix everything afterward If we're going to map the entire physical world, privacy can't just be an afterthought