A woman said an Audi loaner car’s Lytx camera recorded her conversations, sparking privacy concerns over dealership fleet monitoring.
A woman said an Audi loaner car’s Lytx camera recorded her conversations, sparking privacy concerns over dealership fleet monitoring.
Does this also mean they can’t discuss patient information around Alexa or Google Nest, or pretty much any smartphone? It’s really hard to not get your voice recorded and sent to the corponet these days.
Private conversations - doctors, security - can’t be held within earshot of Alexa or its ilk simply because the wake word triggers the sending of audio to America, which violates privacy and sovereignty regs in orgs that pay attention to the rules they all must adhere to and don’t want to be investigated.
This often means phones to be used near those conversations require some disabling of agent trash.
In the case of Alexa or Google Nest, it’s not actually constantly recording or sending stuff to the cloud. Those things have a special wake-word chip, that is trained on one or a few phrases (okay google). That’s fully offline, and only once it detects the wake word does it activate the actual microphone and processing.
If you believe Google and Amazon, that is. Which I surely don’t. But if they were actually constantly recording and sending that data somewhere, security researchers would have noticed odd traffic in their network.
Tl;dr: However they’re supposed to work you can’t trust them.
There’s no “chip.” Alexa, Google assistant, and Siri are always listening. They are looking for the command word, and in the case of Alexa and Google assistant they take at least a thirty second buffer from before the command word for “context.” It was verified several years ago (not sure if it has changed since) that Siri is the most conservative, stripping anything before the command word. All of this is handled in a subroutine that is running as part of the OS for these devices.
There have been documented cases where Alexa and Google devices were recording and uploading 24/7, these were dismissed as “bugs” but they were also only discovered because the users were checking network usage at home. I have not heard of that happening with Siri but that doesn’t mean it hasn’t.
Documented where? I’ve never been able to find any sources for that claim.
This was the one for the Google Home Mini… the fabric on the top was causing some of the devices to constantly record (there is a sensor up there that you could tap to ask a question instead of using the wake word). Once it was discovered Google disabled the tap feature.
I remember hearing about issues with Alexa also constantly recording some years ago, but I don’t recall the specifics of that case enough to be able to easily search for it.