It’s funny because we’ve done this exact testing with the Facebook application on iOS by leaving my friend’s iPhone14 with the screen locked next to Telemundo (a Spanish only public television channel) for 24 hours. (Our primary language is Ukrainian)
The next day, all of their ads were in Spanish.
So I do think additional research is needed for certain, the polling rate might be not as granular as you mentioned, but intermittent anonymous data collection like “primary language” could very likely be done passively with minimal impact on battery life, and it may be permissions-based and operating system dependent.
There is a lot of misinformation on what Facebook is and isn’t doing. And a lot of it is pushing 10 years old.
Facebook has long had features that detect exactly what you’re describing. They aren’t recording it, they are fingerprinting it. The target is any ads and music that is played but it could go beyond that.
This is fundamentally no different than the way a device is passively listening for the “hey, assistant” phrase which just matches a fingerprint.
Anyone who is simply looking for immediate data transfer when this occurs is a fool. There is absolutely no reason it cannot hold the list of known finger prints and add them to otherwise normal requests. The same for anyone looking for cpu spikes; these fingerprints are highly performant and it’s not recording, it’s matching so Facebook can deny all day that they don’t record your conversation and it isn’t a lie because it’s the wrong accusation.
You make me (a skeptic) want to test this in a robust fashion.
Source some foreign-language content offline without carrying/using electronics… record/catalog the ads shown to factory reset Android & iOS devices… let the devices hear the foreign-language content played on an offline system… record the ads shown afterwards. Ensure no other electronics are present.
What else would be needed?
Done in a bulletproof fashion (probably can get some blinding in there too), it would be ProPublica/EFF’s story of the year, and congress would get in on it. Think it could be easily done for a few hundred bucks in about a week. (Thus I’m skeptical of course, such a low barrier to entry relative to the front-page newsworthiness of the scoop.)
It’s much easier for apple to have shared the data that your friend watched Telemundo for 24 hours and thus either has a friend with them that speaks Spanish or is learning Spanish
Or for the Facebook app on their phone to have noticed another app get installed with those details
If you’re not using a smart TV connected to an apple account or an app on the phone to watch Telemundo the only way they could even have that data is if they fucking recorded it using the microphone of your phone. 🤦♂️
Even if not for nefarious reasons, the mic is always listening for the voice activation prompt for when you want it to listen and talk to you.
Ah, so it was viewed on cable or similar? Through a service that likely has a deal to sell/share data about ads being viewed where and how?
Services communicate to each other, thats the entire point of the “your phone isn’t listening to you” thing
Even if not for nefarious reasons, the mic is always listening for the voice activation prompt for when you want it to listen and talk to you.
And that data does not go to anything other than the part dedicated to that. Rather than make a decent argument you’re just kinda showing you don’t know what you’re on about my guy
It makes absolutely no sense for advertising to switch all advertising to Spanish from a single day of recording. This would mean they disregarded ALL of the meta data they had on them. Location, things they visited, pages they visited etc. I’ve been on vacation and spoken a different language for two weeks and it didn’t change the language of my ads. It just makes no sense to do that from a single data point, when all else contradicts them being/speaking Spanish.
It’s funny because we’ve done this exact testing with the Facebook application on iOS by leaving my friend’s iPhone14 with the screen locked next to Telemundo (a Spanish only public television channel) for 24 hours. (Our primary language is Ukrainian)
The next day, all of their ads were in Spanish.
So I do think additional research is needed for certain, the polling rate might be not as granular as you mentioned, but intermittent anonymous data collection like “primary language” could very likely be done passively with minimal impact on battery life, and it may be permissions-based and operating system dependent.
There is a lot of misinformation on what Facebook is and isn’t doing. And a lot of it is pushing 10 years old.
Facebook has long had features that detect exactly what you’re describing. They aren’t recording it, they are fingerprinting it. The target is any ads and music that is played but it could go beyond that.
This is fundamentally no different than the way a device is passively listening for the “hey, assistant” phrase which just matches a fingerprint.
Anyone who is simply looking for immediate data transfer when this occurs is a fool. There is absolutely no reason it cannot hold the list of known finger prints and add them to otherwise normal requests. The same for anyone looking for cpu spikes; these fingerprints are highly performant and it’s not recording, it’s matching so Facebook can deny all day that they don’t record your conversation and it isn’t a lie because it’s the wrong accusation.
You make me (a skeptic) want to test this in a robust fashion.
Source some foreign-language content offline without carrying/using electronics… record/catalog the ads shown to factory reset Android & iOS devices… let the devices hear the foreign-language content played on an offline system… record the ads shown afterwards. Ensure no other electronics are present.
What else would be needed?
Done in a bulletproof fashion (probably can get some blinding in there too), it would be ProPublica/EFF’s story of the year, and congress would get in on it. Think it could be easily done for a few hundred bucks in about a week. (Thus I’m skeptical of course, such a low barrier to entry relative to the front-page newsworthiness of the scoop.)
It’s much easier for apple to have shared the data that your friend watched Telemundo for 24 hours and thus either has a friend with them that speaks Spanish or is learning Spanish
Or for the Facebook app on their phone to have noticed another app get installed with those details
Its not the microphone
If you’re not using a smart TV connected to an apple account or an app on the phone to watch Telemundo the only way they could even have that data is if they fucking recorded it using the microphone of your phone. 🤦♂️
Even if not for nefarious reasons, the mic is always listening for the voice activation prompt for when you want it to listen and talk to you.
Devils advocate here (for clarity im actually on your side)
If youre connected to the same network apple can definitely work out you’re friends
Ah, so it was viewed on cable or similar? Through a service that likely has a deal to sell/share data about ads being viewed where and how?
Services communicate to each other, thats the entire point of the “your phone isn’t listening to you” thing
And that data does not go to anything other than the part dedicated to that. Rather than make a decent argument you’re just kinda showing you don’t know what you’re on about my guy
It makes absolutely no sense for advertising to switch all advertising to Spanish from a single day of recording. This would mean they disregarded ALL of the meta data they had on them. Location, things they visited, pages they visited etc. I’ve been on vacation and spoken a different language for two weeks and it didn’t change the language of my ads. It just makes no sense to do that from a single data point, when all else contradicts them being/speaking Spanish.