Floxk AI is a startup that apparently runs highly suspicious police data collection services. As a fun bonus, it sells this data to private bidders and retains no liability for any mess-ups.
Apparently, it uses a Bluetooth system to relay some of this information. And seems to generally have awful security...
The real scandal seems to be the use of data brokers' data, re-selling/owning the data collected from publicly funded cameras, and also using bluetooth and ostensibly other outdated security.
The scandals you mention have been reported on in depth by 404media, which has led to state and congressional investigation. A Google search will provide technical details about their cameras and the Android/hardware stack that operates them.
I watched the video, and I think the creator of the repo might be concerned about liability. Also, the perturbed plates don't look that interesting, they just look slightly dirty
Aside: this is the third or fourth time this week I've seen a repo/project shared here with some visual output component and no examples shown in the Readme or anywhere and multiple people asking about it. What's with these maintainers missing that key aspect? ¯\_(ツ)_/¯
I've been wondering for a long time about adding a layer of UV reflective material to a license plate (maybe even a spray?), built specifically to either alter the image so a camera sees different numbers/letters, or to block it altogether. With this it wouldn't be obstructing the licence plate at all for human eyes, so it's feasible that you could easily get away with it.
If you see the video, the initial iteration just looks like specs of mud on the plate with the letters+numbers still fully readable. His ultimate goal is a custom plate frame that looks normal to the human eye but can block camera readings.
I can see this as being legal because it's still human readable.
Plate readers aren't expected to be 100% reliable as is (angles, lighting, network goes down, etc.) Plates get dirty, sometimes rusty. Also you can't test your own plate for machine readability against all the different types of systems cops use, so how could you reasonably know the issue is on your end and how to fix it?
Post COVID, where the streets are full of terrible/reckless/dangerous/uninsured drivers with missing or fake (ghost) tags, I have a hard time opposing license plate enforcement.
Floxk AI is a startup that apparently runs highly suspicious police data collection services. As a fun bonus, it sells this data to private bidders and retains no liability for any mess-ups.
Apparently, it uses a Bluetooth system to relay some of this information. And seems to generally have awful security...
Any thoughts? Or further projects on this?