Guys, ten or hundred of thousands security researchers have been going at this for years. Google isn’t secretly listening to you.
These things work with 2 mics, and 2 different circuits. The recording mic is one, while the detection mic is another. The second mic is only capable of pattern matching.
So yeah it’s on but only capable of hashing a 5 second recording and matching it to your voice (this shit works a lot like rsa keys if that’s helpful) to serve as a wake word. Maybe flag a simple response.
All that’s happening is the device heard a loud sound and knows it wasn’t a match or what’s expected.
That’s the neat part. You don’t. There is an entire industry of devs trying to be the guy who conclusively proved all the companies are actually recording you.
It’s far too easy to change the software that drives that. For example, in order to minimize blatant power drain the trigger mic could easily become a switch that activates the main mic only when human voices are detected (or even specific voices). With authoritarian governments on the rise — along with the more than willing corporations backing them — I don’t think a bit of paranoia regarding the possibility is unwarranted.
ETA: Also there’s nothing saying the hardware can’t be updated for newer capabilities without anyone on the outside knowing. It’d be pretty easy to get away with once everyone gets lulled into a false sense of security regarding how they work.
Which is why shit is continuously tested. Guys, billions of dollars goes into this. It’s not hard to find extra data pushed into packets. Far more complex shit is the norm.
Guys, ten or hundred of thousands security researchers have been going at this for years. Google isn’t secretly listening to you.
These things work with 2 mics, and 2 different circuits. The recording mic is one, while the detection mic is another. The second mic is only capable of pattern matching.
So yeah it’s on but only capable of hashing a 5 second recording and matching it to your voice (this shit works a lot like rsa keys if that’s helpful) to serve as a wake word. Maybe flag a simple response.
All that’s happening is the device heard a loud sound and knows it wasn’t a match or what’s expected.
You’ll have to forgive me if I don’t believe that those fuckers are being honest and open with our data.
That’s the neat part. You don’t. There is an entire industry of devs trying to be the guy who conclusively proved all the companies are actually recording you.
It’s far too easy to change the software that drives that. For example, in order to minimize blatant power drain the trigger mic could easily become a switch that activates the main mic only when human voices are detected (or even specific voices). With authoritarian governments on the rise — along with the more than willing corporations backing them — I don’t think a bit of paranoia regarding the possibility is unwarranted.
ETA: Also there’s nothing saying the hardware can’t be updated for newer capabilities without anyone on the outside knowing. It’d be pretty easy to get away with once everyone gets lulled into a false sense of security regarding how they work.
Which is why shit is continuously tested. Guys, billions of dollars goes into this. It’s not hard to find extra data pushed into packets. Far more complex shit is the norm.