AcesFullOfKings
so you're suggesting that it's feasible to use a video stream from the car to read lips by, what, paying out of pocket to send the video stream over the built-in cellular? Possible, sure. Then receiving it at a datacentre and running one of the most computationally intensive algorithms in existence on the video stream, 24/7? Or at least only when I'm using the car? so the datacentre is processing presumably tens or hundreds of thousands of these all at once, so at let's say 5mbps per video that could easily be multiple terrabits per second of bandwidth, consuming megawatts of power in order to spy on people in probably the least efficient and most expensive way imaginable, and all to determine that I said I like coke rather than pepsi so the car company can receive $0.01 in selling that to some ad company?
Is that scenario possible? Yes. Is it happening? I am certain that it's not in any rational company. The same line of reasoning applies to listening to your phone's microphone or camera, except there you'd also notice your phone getting hot and the battery dying in an hour or two.
The GPS thing is feasible. I don't know if they're doing that but they could. As soon as you mention transmitting video feeds or cloud AI you're in conspiracy territory.
||Now, Tesla's on-device AI processing using the driver's power bill to analyse video on-site and only send the tiny results into the cloud.. is very different. ||
Someone could simply use GPS and see that the only place in the area that’s open is the gay bar and infer from that, or even lip reading.
weird, I don't remember doing this
relevant - "YouTube server-side ad injection faq" from Ajayyy, the developer of SponsorBlock
https://gist.github.com/ajayyy/f7b1807e13731c25cef4c2c057d022bc#file-faq-md