“What would it mean for your business if you could target potential clients who are actively discussing their need for your services in their day-to-day conversations? No, it's not a Black Mirror episode—it's Voice Data, and CMG has the capabilities to use it to your business advantage.”
Companies DO analyze what you say to smart speakers, but only after you have said “ok google, siri, alexa, etc.” (or if they mistake something like “ok to go” as “ok google”). I am not aware of a single reputable source claiming smart speakers are always listening.
The reality is that analyzing a constant stream of audio is way less efficient and accurate than simply profiling users based on information such as internet usage, purchase history, political leanings, etc. If you’re interested in online privacy device fingerprinting is a fascinating topic to start understanding how companies can determine exactly who you are based solely on information about your device. Then they use web tracking to determine what your interests are, who you associate with, how you spend your time, what your beliefs are, how you can be influenced, etc.
Your smart speaker isn’t constantly listening because it doesn’t need to. There are far easier ways to build a more accurate profile on you.
It’s literally impossible for them to not be “analyzing” all the sounds they (perhaps briefly) record.
[Sound] --> [Record] --> [Analyze for keyword] --> [Perform keyword action] OR [Delete recording]
Literally all sounds, literally all the time. And we just trust that they delete them and don’t send them “anonymized” to be used for training the audio recognition algorithms or LLMs.
The way that “Hey Alexa” or “Hey Google” works is by, like you said, constantly analysing the sounds they said. However, this is only analyzed locally for the specific phrase, and is stored in a circular buffer of a few seconds so it can keep your whole request in memory. If the phrase is not detected, the buffer is constantly overwritten, and nothing is sent to the server. If the phrase is detected, then the whole request is sent to the server where more advanced voice recognition can be done.
You can very easily monitor the traffic from your smart speaker to see if this is true. So far I’ve seen no evidence that this is no longer the common practice, though I’ll admit to not reading the article, so maybe this has changed recently.
If they were to listen for a set of predefined product-related keywords as well, they could take note of that and send that info inconspicuously to their servers as well without sending any audio recordings. Doesn’t have to be as precise as voice command recognition either, it’s just ad targeting.
Not saying they do that, but I believe they could.
So, you and your friend were talking about a subject you obviously are interested in, likely spend heaps of time online searching about, commenting and following on social media and you’re surprised you got an ad for it? Bonkers.
It’s been published by multiple sources at this point that this happens because of detected proximity. Basically, they know who you hang out with based on where your phones are, and they know the entire search history of everyone you interact with. Based on this, they can build models to detect how likely you are to be interested in something your friend has looked at before.
Yup. For companies it’s much safer to connect the dots with the giant amount of available metadata in the background than risk facing a huge backlash when people analyze what data you’re actively collecting.
Which is why people need to call out the tracking that’s actually happening in the real world a lot more, because I don’t really want my search-history leaked by proxy to people in my proximity either.
Companies DO analyze what you say to smart speakers, but only after you have said “ok google, siri, alexa, etc.” (or if they mistake something like “ok to go” as “ok google”). I am not aware of a single reputable source claiming smart speakers are always listening.
The reality is that analyzing a constant stream of audio is way less efficient and accurate than simply profiling users based on information such as internet usage, purchase history, political leanings, etc. If you’re interested in online privacy device fingerprinting is a fascinating topic to start understanding how companies can determine exactly who you are based solely on information about your device. Then they use web tracking to determine what your interests are, who you associate with, how you spend your time, what your beliefs are, how you can be influenced, etc.
Your smart speaker isn’t constantly listening because it doesn’t need to. There are far easier ways to build a more accurate profile on you.
A recent study found these devices incorrectly activate like 80 times per day on average
Link?
On mobile, can’t find the recent one based on conversation that was floating around lemmy recently.
This one finds high levels of inconsistent misactivation from TV shows. Some shows caused more than 4 misactivations per hour (a rate of more than 80 per day) https://moniotrlab.khoury.northeastern.edu/publications/smart-speakers-study-pets20/
deleted by creator
It’s literally impossible for them to not be “analyzing” all the sounds they (perhaps briefly) record.
[Sound] --> [Record] --> [Analyze for keyword] --> [Perform keyword action] OR [Delete recording]
Literally all sounds, literally all the time. And we just trust that they delete them and don’t send them “anonymized” to be used for training the audio recognition algorithms or LLMs.
It is possible to analyze the traffic leaving these devices, and AFAIK it hasn’t been shown that they are doing this.
The way that “Hey Alexa” or “Hey Google” works is by, like you said, constantly analysing the sounds they said. However, this is only analyzed locally for the specific phrase, and is stored in a circular buffer of a few seconds so it can keep your whole request in memory. If the phrase is not detected, the buffer is constantly overwritten, and nothing is sent to the server. If the phrase is detected, then the whole request is sent to the server where more advanced voice recognition can be done.
You can very easily monitor the traffic from your smart speaker to see if this is true. So far I’ve seen no evidence that this is no longer the common practice, though I’ll admit to not reading the article, so maybe this has changed recently.
If they were to listen for a set of predefined product-related keywords as well, they could take note of that and send that info inconspicuously to their servers as well without sending any audio recordings. Doesn’t have to be as precise as voice command recognition either, it’s just ad targeting.
Not saying they do that, but I believe they could.
deleted by creator
So, you and your friend were talking about a subject you obviously are interested in, likely spend heaps of time online searching about, commenting and following on social media and you’re surprised you got an ad for it? Bonkers.
It’s been published by multiple sources at this point that this happens because of detected proximity. Basically, they know who you hang out with based on where your phones are, and they know the entire search history of everyone you interact with. Based on this, they can build models to detect how likely you are to be interested in something your friend has looked at before.
Yup. For companies it’s much safer to connect the dots with the giant amount of available metadata in the background than risk facing a huge backlash when people analyze what data you’re actively collecting.
Which is why people need to call out the tracking that’s actually happening in the real world a lot more, because I don’t really want my search-history leaked by proxy to people in my proximity either.