I still don’t think companies serve you ads based on spying through your microphone
(simonwillison.net)
from Olap@lemmy.world to technology@lemmy.world on 02 Jan 2025 23:53
https://lemmy.world/post/23823698
from Olap@lemmy.world to technology@lemmy.world on 02 Jan 2025 23:53
https://lemmy.world/post/23823698
Not my title! I do think we are being listened to. And location tracked. And it’s being passed on to advertisers. Is it apple though? Probably not is my take away from this article, but I don’t trust plenty of others, and apple still does
threaded - newest
Ads aren’t why you should be concerned about apps w/ microphone access…
Where exactly are you getting the idea that this belief is widespread?
Not my title, as I already said. But anecdata backs this up ime. Go ask your parents for a giggle, see what they say
I have heard it repeated several times. It’s based on how virtual assistants are allowed to listen over your mic for keywords, applications like Facebook requesting full microphone access, and people with stories of getting ads for things after having a conversation about the same.
The third could be a form of recency bias; I just learned about this, and now I see it everywhere. Also, it’s easy to know who is in your circle, and items you recently searched could be advertised to your friends. I saw this by getting sudden ads for handguns after getting an Amazon link from my gun crazy friend.
Like people allowing google Gemini to always listen for permanent background Shazaam functionality and and having surprised Pikachu faces lol.
as I posted this, found this Apple law suit for Snooping Siri post on my feed lol.
Ugh and Google fingerprinting free for all Feb 2025
me, a GrapheneOS user 😈🙈🙉🙊
I saw it happen about 3 years ago. I mentioned Omega watches to my buddy who had somehow lived to 50 w/o ever hearing the name. Later that same day there was a Facebook ad for Omega on his Samsung phone.
Yep me too. Wife was adamant her phone was spying on her so we decided to test it by talking about Lexus cars (having chosen cars because at the time she was getting no car ads anywhere, had no interest in cars, and she had never heard of the brand before so certainly hadn’t searched for it). A few hours later, her Facebook feed was full of Lexus ads. 100%.
On the one hand: en.wikipedia.org/wiki/Frequency_illusion
On the other: nypost.com/…/marketing-firm-spies-on-you-through-…
I’ve heard many folks suspect it. It’s a widespread, if weakly substantiated concern.
Apps listening to your mic to give you targeted ads is an urban legend. There’s tools to see which apps listen to you and there isn’t any evidence that any of the popular stuff ever open the microphone (unless you’re in a call or something). If you’re too worried about it, you can always turn off the mic permission for the app.
The ads are actually coming from other ways of tracking you like browser fingerprinting to follow what things you browse and build a profile on what you like/are interested in.
See also EFF’s article on it: digitalrightsbytes.org/…/is-my-phone-listening-to…
agreed. online tracking is so good it just seems like they’re listening to you.
The ones serving up the ads aren’t even the ones listening. They’re buying collated data from many different sources, then their algorithm matches your interests with one of the products they’re contracted to sell. Next thing you know you’re looking at a Rolex ad because you zoomed in on someone’s watch on their Instagram post.
I’m not so sure. When my partner and I were on a road trip we had android auto connected and were singing along to songs we listened to via Spotify. At some point though, when I tried to fiddle with some settings the connection between the car and the smartphone bugged out and while trying to fix it we suddenly heard his voice being played back on the speakers “whispering” some lyrics he had sung 30 to 60 minutes earlier.
I put whispered in quotes because he certainly didn’t whisper those lyrics and I recalled the moment he sang them quite clearly. Beside his singing and the music playing there were no other sounds at that time.
My best guess is that he was actually recorded while singing and something was stripping all the background noises and music to make his speech more clear for speech to text analysis. It was creepy as fuck.
We both work in IT and I truly have no other idea what this could have been given the circumstances. He said there is actually a company that provides a framework that listens to, records and analyses whatever is spoken near smartphone microphones and all the big tech players like Google are using it. I don’t remember the name though. Would have to ask him.
That’s the most unlikely story I’ve heard in a minute… Even assuming there’re some deep rooted kernel level shenanigans, which no one has found yet, how would you fiddling with some settings expose that?
Probably just got a dropped call, and it resumed the playlist in shuffle, I’ve had it happen where the music comes out as if in a phone call (messes up frequencies) for a few seconds before it goes back to normal. Occam’s razor and all
Believe what you want, I don’t really care. The whole connection bugged out and the car’s infotainment system including android auto became unresponsive. There was no call, it wasn’t shuffle and it was definitely his voice, not the music playing. Especially since there was only the whispered singing. No other instruments at all.
I believe it was sunspots that caused bit flips on the phone CPU and regurgitated data from the L5 cache. /s
Jfc, finally some sanity in this thread. Thank you. You’d think a bunch of supposed computer nerds would have done a fucking experiment before going off on some anecdotal bullshit.
Why wouldn’t you think this? There is no system in place for monitoring those companies, nor is there any type of punishment for if they were to be proven to be doing so. While on the other hand, there are piles of money to be made from advertisers for allowing exactly that to happen.
I’ve personally had things come up as being advertised to me after being NEAR people talking about those items, and I have seen several videos where people show this effect in action.
Frequency illusion is real, but is not reliable enough to repeat over and over, back to back, unlike the advertising.
When, ever, have the capitalist companies prioritized morality over money? Never.
While there is no system for monitoring the companies, experts can reverse engineer the apps and debug the devices. Thusfar, experts who have done this have found no evidence of these types of activities. All the evidence is anecdotal. I believe if this was a widespread practice, evidence would have been uncovered by now and we would have been reported on widely.
The implication here is really scarier than if they were listening to our conversations. It means they do not need to listen to our conversations. The telemetry they already have is so good that in many cases they know what you will say with such high degrees of accuracy that people assumed that they had to be spying on their conversations.
Either way, we need to demand an end to this unprecedented mass surveillance.
battery life would fall through the floor if they did spy
Yeah, there’s just more effective methods to get essentially the same data.
That was once true, but I am now very skeptical of that with on-device processing that can log key words and send them without using much data or power.
and how are key words decided?
AdSense
That's a ton of words
Yes this is how algorithms work
I don’t know what you mean. Every ad campaign has its own keywords. They add up to a ton.
devset.ai/…/revolutionizing-adsense-harnessing-th…
Gemini is on every pixel 7-9
I don’t see how this clearly AI-generated article about how we’ll have 1-on-1 conversations with ads has to do with what we’re talking about.
Ok here’s from the google dev blog
blog.google/…/put-google-ai-to-work-with-search-a…
Here’s how a model can store a localized profile from adsense and learn on it
researchgate.net/…/316821039_Convolutional_Dictio…
Here’s how google tensor chips are literally built for that workflow to the detriment of other performance.
androidauthority.com/google-tensor-g4-explained-e…
Google has an adsemse profile on you the user, that profile is added to by metadata from apps and sensors on the phone; then offloaded to google cloud servers when the phone is charging. No input from the user required.
Here’s a GrapheneOS security feature to prevent persistence and breaks the above workflow.
grapheneos.org/features#anti-persistence
I have an honest question. How much android kernel development or tensor chip work have you done?
That Google blog also says the same thing except it's written by a human. I'm not disputing that AI can process audio data into ad statistics; I'm disputing that audio data is constantly recorded and sent.
I interpreted Farts as saying that the device listens to key ad words just like it listens for "Hey Siri", and I asked how it decides which words to listen to. Each ad campaign has their own keywords, and if you want to personalize, you'll have to listen to all of the words from every campaign, which would be equivalent to listening to everything and would severely degrade performance.
Reread my response to you. It’s not a stream, it’s locally held as tokens and then streamed over wifi while charging. I don’t see any sources posted by you. I think we’re wasting each other’s time.
And the tokens would have to be analyzed from audio, which has to be recorded first. Held temporarily or not, it's mostly the same.
A phone reacting to “ok Google” or the equivalent for the other assistants already requires to listen to what you’re saying - doesn’t seem to affect battery life all that much.
Passively waiting until a specific pattern of pressure waves impacts a sensor takes way less power than recording and transmitting the voice data
Instagram showed me an ad for a medical condition I only discussed out loud, in person, in my doctors office.
Instagram was immediately uninstalled that day.
Other methods of data collection can be scarily effective. Stores have identified people were pregnant before they knew.
Very likely they identified you as someone that could have that condition, and you noticing the ads after talking to your doctor is a form of recency bias.
You can collect almost all the same data from traditional surveillance methods. Collecting and processing mocrophone data just isn’t effective enough to make up for the massively increased costs from processing it.
It displayed the ad before I could get home and research it. It had only been discussed out loud and in person.
There’s always other signs.
Did you connect to the clinic’s WiFi?
Just being near their WiFi is enough.
Yes, I was on the wifi before the appointment.
This is why. Not because your phone is listening to you.
The wifi was not related to the medical issues, the symptoms, or the specific clinical term mentioned by the doctor.
But it’s reasonable to believe other people would have searched the same topic on that network
Edit: or even that they looked it up o their own network following connecting to the clinic’s.
As much as I logically know this to be the case, especially now that Android and iOS indicate when things like the mic are active… My brain still wants to reject it because it is just too coincidental.
I do not trust mic switches however, unless someone can provide proof that it physically disconnects the circuit to that microphone, it can be bypassed somewhere and there’s no reason to trust the manufacturer.
I hate to add to the conspiracy, but I know my eye doctor uses a 3rd party which has sections of their hipaa privacy acceptance which allows them to use your info to sell you ads if you don’t decline. Phreesia, is the 3rd party company. Now add the other apps that track your location… time spent there…
and I know my grocery store does the same when you use the discounts. and worse, they have facial recognition so I can’t even opt out (kroger).
Your issue was likely a combo of that.
Can’t be facial-recognized if you always wear an N95 mask in stores ;)
A random person recognized Luigi in just a couple days, and he was wearing a mask in all the video footage released.
Modern recognition systems can scan footage a lot faster than humans. Many modern systems don’t just use facial recognition but other factors like general height, walking gait, stride length, etc. to make more accurate recognitions.
Wasn’t his face revealed in cab footage where he’d taken his mask off within like, a day?
I think there is also probably a difference of scope in what is leveraged when Kroger is trying to get your facial pattern while you’re in the store to track where you go and get more data to sell advertisers vs like, the lengths gone to by the state in order to catch someone who shot a rich person dead in the street.
Sorry, but facial recognition software has basically caught up. I would not rely on a mask to prevent me being recognised today:
https://privacyinternational.org/news-analysis/4511/can-covid-19-face-mask-protect-you-facial-recognition-technology-too
https://www.ft.com/content/42415608-340c-4c0a-8c93-f22cdd4cc2d6
https://www.techtimes.com/articles/304431/20240508/new-software-shows-promise-facial-recognition-underneath-mask.htm
It must be very situational on when it works, because the FBI has specifically cited a face mask as a reason that it still hasn’t caught whoever left a bunch of bombs laying around on Jan 6.
Interesting. I can imagine a scenario where the resolution of CCTV is low enough that a mask would impede recognition in that instance. It's definitely not something I would want to rely on, though.
Your age group, sex, location, profession/industry, income estimation - you can assume they have this data.
That + a few data points that could be tracked by apps or websites:
Cross some of that, personal info, and ads of treatments for conditions of X.
They don’t need to listen to your mic.
That said, if it’s a fairly common condition, it might be the case you were presented the ad before and never noticed it.
None of those data points apply. It was nothing I had searched for or spoken to anyone until I saw the doctor that day and the Instagram ad was present by the time I had driven home, specifically mentioning the clinical term mentioned by the doctor.
It wasn’t even the stated reason for my visit, it was an afterthought at the end of the appointment… “Oh yeah, as long as I’m here, what is this…?”
Believe it, don't believe it, some people believe the earth is flat, and 5G gives you cancer
404 Media previously reported Cox Media Group (CMG) was advertising a service that claimed to target ads based on what potential customers said near device microphones. Now, here is the pitch deck CMG sent to prospective companies. Google has kicked CMG off its Partner Program in response.
We live in an age where the voice can be processed locally on the phone (we’ve had on-device speech-to-text since the late 90s…), and it’s already listening for a wake word, meaning mic is always hot. It doesn’t need to be streamed and use bandwidth; it can fire off 4K of JSON every few hours and relay more than enough information.
Just program whole dictionary of key phrases and scan the wake word buffer like you are already doing. Easy, stealthy, encrypted. Every voice assistant from a major tech company could (and likely IS) doing this.
This also provides ample opportunity for domestic (or even foreign!) spying my state actors, too.
Yes Apple.
Apple is the one who got caught so far
If you think Google, Amazon, Meta, Microsoft, Samsung et al aren’t doing this, I’ve got a bridge to sell you.
Ok? I didn't say differently. OP said
Did yall read the article?
I’ll buy-t. What did Apple get caught doing that these other companies haven’t got caught doing?
Edit: Oh, the Siri settlement. The article linked argues against the claim of it being used for advertising, though.
I don’t either hut the alternative is much worst in my opinion. It would mean the algorithms are so advanced they are predicting conversations instead of listening to them.
…time.com/…/how-target-knew-a-high-school-girl-wa…
Nah.
kdnuggets.com/…/target-predict-teen-pregnancy-ins…
ITT:
People saying “They already use every other bit of data they can access, why do you naive optimists think they wouldn’t use the most obvious one?”
vs.
People saying “They already use every other bit of data they can access, why do you naive optimists think they would need to use the most expensive one?”
it’s effective, timely, accurate, and profitable.
ofc they’re gonna use the audio, too; where and when possible.
Aw jeez not this again.
I’m not saying it’s completely 100% not possible and has never happened in the history of human technology, but the situation is not as ubiquitous as most people seem to think it is.
Don’t get me wrong, collecting and inferring personal information is happening on an epic and ubiquitous scale these days, but for the most part, it’s not the microphones on your devices that are doing the data collection.
Pretty much all my older relatives are completely convinced their phones are listening to their day to day conversations and serving up ads based on those conversations. One of them came to visit me for a week over the summer. One night we had been talking about having asparagus for dinner, and as evidence that their phone was listening to us, the next day they showed me that their news feed was filled with asparagus recipes. Another night, we were talking about one of their medical conditions and the drugs they were taking, and the next day they showed me that they got notifications about a prescription drug for that condition. On another day, we had been talking about a specific actor’s filmography and all their movies that we liked, the next day their streaming video app was suggesting a bunch of content from that actor.
I can understand why this seemed pretty convincing that our phones were listening to us, but consider the simpler explanation.
I live in a rural area where there’s not good cellular reception, so for the most part, our phones are connected via wifi to the same internet connection. Essentially, every device on the property has the same external IP address. So, when I looked up asparagus recipes on my laptop later that night because I wanted to surprise my relative with that specific dish, and when I Googled the prescription medication the relative was taking to see what the side effects where, and when I looked up that actor on IMBD to see what all movies they’d been in, that pretty much gave all the advertisers all the information they needed to start targeting ads and recommendations to folks sharing the same IP address.
Occam’s Razor being what it is, I assume that’s how things went down versus all our conversations being constantly recorded and uploaded to the net to be interpreted and used for the purposes of serving ads.
I talk to my father on the phone.
We finish.
I receive ads for a very specific thing that we talked about that I’ve never ever looked up.
Same thing with my therapist.
We talk. I receive highly specific ads.
It can always be explained by something else. Recency bias being a big one. It’s very possible you saw an ad yesterday as well, but didn’t notice you saw it because you haven’t talked about that item. Talk about it today, see the same ad, and now you think you’re being listened to.
It’s very possible your father googled something after hanging up the phone. There are endless ways they can connect you to knowing your father.
It’s possible someone on the same wifi network as you or your father overheard the conversation and looked it up.
All of these are far more likely than everything you say and do being recorded without anyone ever finding any definitive proof.
There have been ad agencies who have claimed to be able to do this, so they would if they could. www.404media.co/heres-the-pitch-deck-for-active-l…
Just because an ad agency makes a claim doesn’t mean it’s true or doable.
yeah it does
Right
Except I’ve had experiences that aren’t explainable by alm this:
Discussing a random, never-thought-of-before idea with a friend, in the car. Neither of us had ever thought of this thing before (honestly don’t recall now what it was). Discussed it for 2 minutes, then moved on.
Later we’re both seeing related ads, yet neither of us searched for anything.
And it was something way out of left field for both of us, that neither of us had ever thought of before. The related ads were so jarring that we both told each other about it.
Oh, and my phone was rooted, de-googled (lineage), with heavy restrictions for the apps, no social media (I still don’t have any accounts with any of them, except here), etc. The other phone was an iPhone.
The scary thought here isn’t that they’re actively listening in.
It’s that they know enough about you to know that something will be of interest to you before you even realise it yourself…
If it’s a never before thought of idea how can there possibly be ads for it? No one has ever thought of it and the product doesn’t exist right?
Like I said, there are always other ways. Maybe you searched something related the day before and don’t remember. Maybe your friend did.
Ever had a free free flowing conversation for hours?
You talk about things you wouldn’t possibly think of alone because of the other person.
Im located in Germany ne I guess many of you people are have not yet been victim of this, else you would agree with me.
Perhaps they track who you talk to and show you ads that are relevant to those people, or their best guess based on two profiles.
I don’t think there’s a data center out there with a live audio stream of literally billions of always-on devices 24/7/365.
Perhaps there’s some local processing first, but devices have permissions for apps, and lights that indicate the mic/camera is in use.
I figure someone would have figured it out by now (reverse engineering, decompiling code), or someone from Google/Apple/Samsung would have leaked it if it were true. Think of the number of people required to keep this secret.
And yet my android phone is able to detect what song is playing 24x7 without being a noticeable drain on the battery or using extra data. Doesn’t seem far fetched to be able to do keyword spotting under the same constraints.
Here’s one example of a company getting caught: m.economictimes.com/news/…/113071827.cms
What is this referring to? Spotify?
support.google.com/pixelphone/answer/7535326
Oh like Shazam
On my android it’s called “now playing” and it’s a feature of the os. It even recognizes live cover versions as the song they’re covering sometimes. Claims to do it all local-only without sending data, just always listening and seeing what it recognizes.
Granted, things are getting to the point that it may very well be possible. But these kinds of claims have been around for over a decade, and today my voice control devices still fail to understand me rather regularly.
Not to mention, a song is usually extremely easy to pick out. In a loud bar with background music, your brain tends to pick up the beat and start grooving along with it, even if you don’t realize it most of the time. Compared to the person sitting across the table trying to yell things to you, and you having to resort to lip reading and guesswork.
It listens to music being played on other devices? I haven’t seen it on my phone.
Or live music played by musicians, in this case.
Maybe it is only for Pixel phones. It’s hard for me to differentiate between pixel features and stock android recently.
<img alt="" src="https://lemmy.blahaj.zone/pictrs/image/d62f92d1-3e3a-4f79-82ae-ca06d232fb08.webp">
No. With my therapist I talk randomly about so many things. It’s impossible to predict or calculate.
Also I talk to my therapist in person while the phones in my pocket or table.
Edit: also have you ever had a natural flowing conversation? It’s not predictable because you can possibly talk about everything and anything.
So I’m hearing that you don’t use an adblocker.
.
Only on Spotify, which is another scandal
Timely. apnews.com/…/apple-siri-iphone-lawsuit-settlement…
On Android, I have the mic, location, and camera blocked via the pulldown tiles menu. I turn them on when needed. The OS and some apps like to bitch about this sometimes but it seems to be working ok.
My iphone does not offer these blanket blocking options. It’s a work phone, so I just leave it off unless I need it.
So one of the devices allegedly grabbing keywords from heard conversations, you’d trust with a software based toggle?
I’d only trust hardware toggle.
You really trust that the button does what it says it does though??!??
You mean a hardware toggle? Because those can be double checked with some electronic knowledge.
And a software toggle can be checked with some software knowledge if the source is available.
Although chances of code being available are quite low for most mobile devices, not to mention software takes many many hours longer and you’ll have to make sure your build matches the code (which means manually checking the code of every update).
Question for you: Do you manually check every hardware toggle for correctness?
Weird question. A better question would be how hard it is to check. It’s pretty easy. It only requires a multimeter and basic knowledge of electronics. Opening a phone isn’t hard either. You can easily use a clothes iron with a rag in front of it.
Never mind, I guess? I ask the questions I’d like answers to. You seem to answer questions you’d like to be asked instead of the ones asked. Weird behavior.
I ask because the ease of which something can be done is worthless if nobody does it.
How would a lack of people using hardware toggles even make it worthless? I don’t need that answered. It doesn’t.
I am guessing you use a software toggle and are trying to defend your use of it. Which shouldn’t be directed towards me. You trust it? That’s great. I’m glad for you. I explicitly said I only trust hardware toggles and explained why I wouldn’t easily trust a software one. What I do is of no concern to you.
So next time just share your opinion instead of asking a question you don’t need answered. Which I obviously noticed. Don’t beat around the bush and just cut to the chase.
You misunderstood/I wasn’t clear. I was referring to the ease of which you can verify that the hardware does what it is supposed to do. If you can verify it easily, but don’t, the ease has no meaning.
So… Do you verify all/most of your hardware toggles? Personally?
And no, I don’t use software toggles. I don’t use any toggles on my phone, except Bluetooth to save battery. 🤷♂️ You guessed wrong. Maybe don’t be so smug about trying to “read” someone. It ends up being super cringe when it’s wrong. 🫣
I don’t think you understand how electronics work. Maybe if you have an ugly plastic see trough toggle, but I don’t see that happening any time soon.
I also don’t understand why you are asking me questions because you’ve failed to explain something, as in some ploy to make me the bad guy here.
:') You’re a joke.
I explicitly used the word ‘guess’ to not be smug. As in, the following part is merely a guess. But hey, you can keep making it personal, but it just shows who is “cringe” here.
Now get blocked.
😂 what a tool
Exactly the type of response I was expecting lol.
Yes boy, please block me. Please do.
At this point, it’s the best I can do. I have tape on the selfie camera too. I guess I could bust this pixel 7a open and add physical switches to the cameras and mike, but I’m not real confident in my ability to pull that off. Maybe you can tell me how to do that.
I guess a software based toggle is ok for downloaded third party apps on the phone. Just not anything by Google.
And no I am only saying what I trust. I sadly have no toggle. Making an actual hardware toggle outside of the factory line is pretty hard. But disabling them on your device is pretty easy. You can easily make your USB port power only, unplug your camera and destroy the mic.
If you use Graphene on the Pixel however, I think you’d be safe from anything Google to be honest.
Two ways to process voice, on device or on server. Device-based solutions either are very basic and just detect differences between words or need training data based on your voice or they need lots of processing power for more generalized voice recognition. So is your battery draining and phone is often hot because an app is keeping the mic on and keeping the phone from slowing the processor? Other option is to stream the data to the server. This would also increase battery usage as the phone can’t sleep, but might not be as noticeable, but more evident would be your phone using a lot more bandwidth than is reasonable while you aren’t actively using it.
.
Have you considered carbon monoxide poisoning? Maybe the disoriented you is snitching on you to advertising companies (/s just in case it is not clear).
You are in the same geolocation as other people and they are searching for the stuff you’re talking about. Try whispering to your phone alone in a closet.
I’m not ready for that kind of intimacy.
.
Do you connect to company WiFi?
.
Do you use a VPN for ALL company communications?
.
Not surprising. Samsung is one of the companies most openly spying and breaking laws. They’ve also been caught spying on their customers’ tv microphones.
I don’t think that most of the big tech companies are listening to your microphone (I’m not ruling it out entirely, and I’m certainly there are some smaller sketchier companies that are doing it)
But I think most of the time most of the time they don’t need to
They know what ads you’ve seen on your phone/computer, what you’ve been googling, the websites you’ve visited, where you’ve used your credit card, what shows and movies you watch, and where you’ve been (from gps locations, or from what wifi networks and Bluetooth devices you’ve been near or connected to) and what ads, playlists, stores, products, etc. you were exposed to while you were there, and of course who you talk to and all of that same information about those people.
That’s all going to influence the things you think and talk about, they probably have a pretty good idea what kind of conversations you’re going to have well before you do.
And don’t get me wrong, that’s creepy as fuck.
I think most of it comes down to people not even realizing how much data about ourselves we put out there and all of the ways it can be collected and used to build a profile about you.
And honestly I think they can probably get better data from that most of the time than from trying to filter out background noise and make sense of what you’re talking about through your microphone.
Few hours later on timeline:
lemmy.world/post/23832012 (Apple randomly spying using microphone)
It kind of sounds like that article is about the “hey Siri” feature getting activated accidentally, like if it picks up something that sounds similar to the trigger phrase and starts recording
Which is still a big security/privacy issue, but not exactly the same as if they’re just turning the microphone on whenever they want to listen to you and serve you ads
But most likely they use all of those recordings accidental or not to serve you targeted ads, like Google does.
not really the same thing but it could happen to anyone: reuters.com/…/apple-pay-95-million-settle-siri-pr…
They need to demand a code review to find out who set up some of these triggers. Especially the zipper sound. Send that creep to prison.
Calm Down—Your Phone Isn’t Listening to Your Conversations. It’s Just Tracking Everything You Type, Every App You Use, Every Website You Visit, and Everywhere You Go in the Physical World
phew
Except it is also listening. This was a minor scandal back in September. I believe Cox media has since been dropped by Facebook and Google and such, but it happened.
news.itsfoss.com/ad-company-listening-to-micropho…
That article covers a pitch deck by an ad agency with absolutely zero detail of how it works.
If this is happening, it should be easy to test.
variety.com/…/active-listening-marketers-smartpho…
Apps still need mic permissions to do so. Many Android ROMs include notices when the mic is being used, it would be very easy to tell if an app was actually doing this.
I once worked in a charity providing mental health services to people without insurance, or who wanted to not have their insurance record the service for whatever reasons.
I once had a homeless man that I would see regularly. We set up each appointment at the end of the preceding appointment, because the only other way to get a hold of this person would be to call the fast food place he worked at, during his work hours, which weren’t consistent. This man did not own a phone, or any other electronic device. His facebook, and all of his online activity was done at his local library. I emphasize this because I need it to be stressed that there was no way any algorithm could connect his location to mine. There was no way for a system to recognize that his device was near mine, because he did not have a device. There was no way for any of his online habits to be algorithmically connected to mine, at all.
One session, we’re speaking. The only devices in our small, sound proofed room, were my cell phone, a digital clock not connected to any system, and a digital camera, turned off, and also not connected to any system. He mentions that he’s been contacted by someone who wants him to move to the Phillipines. We briefly discuss flights and work in the Phillipines. Then we move on to other things, yadda yadda, end session.
By the end of the day, I’m getting ads on Facebook for flights to the Phillipines. Freaked me the fuck out because those sessions are HIPAA protected. From then on I kept my phone turned off, and in a completely different room in our building than any of my sessions with any patient. Never ever had it happen again.
I’ve had a very similar experience.
Once discussed something, out of the blue, something I’ve never been curious about in my life, in the car, with a friend who also has never thought about the same thing.
Hours later we’re both seeing related ads.
Now, I get that the amount of data required for such analysis is supposedly outside the bounds of what phones can do. But I can’t see any other explanation. Neither of us ever searched anything in this subject, we talked about doe a couple minutes and moved on, never doing anything about it. We have very different interests, too.
Difficult to judge. Could be confirmation bias, as well. Meaning you got ads for flight befores. But you were not paying attention to them at that point. Which changed after your session and now you think these are connected. (Or you looked something up about that location and that kicked it off.)
These are the usual findings in the rare cases people are able to trace it back and they write some article or podcast about it. Mainly confirmation bias. And once you interact with one ad that got you taken aback, you're trapped.
Doesn't rule out other possibilities, though. I guess what I'm trying to say is, this counts more as anecdotal evidence. And we have plenty stories like this. It's not enough to infer anything. More a reminder to investigate some more.
And yes, it's good practice to keep your phone someplace else when you're having protected/confidential conversations. Smartphones are very complex and they certainly have the potential to spy on you. In fact we know a lot of the apps and computer code is meant to analzye your behaviour and transfer that information to third parties.
How many anecdotal stories before it becomes data? If hundreds of people are saying that this happens and there’s no other explanation? Thousands? How many things can be written off as “Oh, something you don’t understand is happening, even if we can rule out basically everything.” ?
With the scientific method and anecdotal evidence: kind of never. It's illegitimate to draw that conclusion, this way.
You got to dig down to the facts. Or we can just tell the fact that a lot of people feel that way. And I mean "confirmation bias" is a very good explanation. We also have thousands of people believe in esoterics, homeopathy etc. The mechanics of psychology are well-understood. And it's kind of the reason why we invented science in the first place. Because we found things aren't always as they seem. And there are a lot of dynamics to factor in.
If we want to get to the truth, we have to do a proper study. I'm not an expert on this, so I don't know if we got to that, yet. I know people have demonstrated this is technically possible. But as far as I'm aware people have also taken apart a few of the major apps like Facebook etc, logged the traffic and couldn't find anything that uses microphone data to do targeted advertising.
Conclusion: It's either not there, or we missed it.
Now I want to do some kind of experiment where I speak things into my phone and see what happens. It still seems too much to be coincidental.
Playing devil’s advocate here - we know voice information is being sent back to both Google and Apple, if the analysis were done server side dissembling apps isn’t going to show us anything we don’t already know.
That's kind of the important question... We know it sends audio on request, if you trigger it somehow. But does it transmit anything clandestinely in the background? And does it suppress any microphone icons from showing up? I believe that's where disassembly and sniffing network traffic come into play.
There are billions of smartphones out there. Thousands of people getting ads relevant to what they just discussed is normal. And it’s not just about the number of stories. It’s also about how unscientific these reports are as well. If you want to come up with actually useful evidence you would have to test this multiple times to prove it’s not random and you would also have to objectively measure the effect. You need to show a significant increase in the probability of getting a relevant ad, which in turn means you need to know what the baseline probability of getting one is (when the phone has not been allowed to spy on you).
All that being said, I don’t think proving that smartphones spy on us is all that useful. The fact that it can happen very easily is already a problem. Security and privacy are protected when we design systematic solutions that prevent abuse. They are not protected in unregulated systems where we might sometimes prove abuse has happened after the fact. There’s plenty wrong with a modern smartphone regardless of whether it happens to be spying on you right now.
Btw, I think it's pretty much accepted fact that smartphones do spy on everyone. It's the main business model of any big tech company. Google, Meta... They definitely have algorithms to tailor their targeted ads to someones personal profile. And per default they look at what you're doing online all day. Keep track of your location if they can... The one thing that's unclear is whether they use the microphone and also listen to your offline conversations. My main point being: Listening in with the microphone isn't that far off. If you feel uncomfortable with that, you might want to re-consider a few other things as well.
Thanks for the heads up. I am aware of the spying issues with smartphones (and any way you access the internet really). This is part of the reason why I don’t think proving the unauthorized use of the microphone to spy is really important and why we need systemic solutions to prevent abuse in any case.
Can there really be an objective measurement? You should think first thing data harvesters would implement is a sort of cloak, to erase any traces of what’s going on. Think Dieselgate, but more sophisticated. E.g. phone detects it’s being tested the way you described, or is in the hands of a state attorney or whatever, the recording/ forwarding/ prcoessing of data stops.
That’s only really feasible for phones they knowingly send to regulators. The phone would have no practical way of knowing that I’m having staged conversations around it and keeping track of the ads I see.
But even if you’re right, that doesn’t change the fact that a lack of objective measurement means all these stories are unreliable.
I think that no amount of anecdotal evidence would be enough. For a very long times doctors had anecdotal evidence that bloodletting saved patient, yet they were fooled by their bias. I’m not saying advertising isn’t spying on our microphones, I don’t know, it might be. But it doesn’t seem very plausible to me: the amount of processing necessary, and the amount of network seems way too high. Also, voice recognition is still not great currently, it was even worse years ago.
The plural of anecdote is not data
Yes, this is the common statement I am referencing.
What made you bring up the Philippines in the first place? Even if you have not been served ads before then, or the other guy. Someone either of you have interacted with could’ve done who brought up the Philippines to you or them.
And because there’s an ongoing campaign in your area, eventually you’ll get one of them ads too.
As I said in the original post, the client was contacted by someone over social media about moving to the Phillipines for work. It turned out to be a scam. Nobody else I interacted with made any mention of the Phillipines to me.
Yeah but that scam may have been going around the area elsewhere and had caused a spike of searches in your area so the add companies programmatically fill in what they see as an area with potential leads with ads.
Great story.
Even if anecdotal fuck all of that better safe than sorry.
My dad use to say that Facebook listened to him back in the 2010s. We blew him off as conspiracy nut.
He would say diamond ring diamond ring diamond ring and then all his ads would change next day. We blew him off as conspriatorial and now the algorithm is common knowledge.
Who knows. Scary.
Same here. Confidential discussion with lawyer/ doctor/ pharmacist, get extremely relevant ads at once. Therefore, I made it a habit to completely turn off my phone before entering such situations, and, if I can, put it in a switched-off microwave or some other Faraday cage structure, Snowden-style.
The comments here show the real problem, adverts dont have to say why they’ve been selected.
All online ads should have to say which filters they matched to advertise to you. The advertising in most cases now is centralised into Google or Facebook, this is absolutely technically possible.
According to the Signal foundation, the reverse is true. They claim they got banned for revealing that info.
signal.org/…/the-instagram-ads-you-will-never-see…
Wow.
This gives me that gut uneasy feeling. Those grabs are hyper specific examples.
Sharing data with the plebs is the true crime.
They are absolutely listening. This is very easy to test. Speak around a Google or Amazon or Apple device, and start talking about things you would never buy, never need, have never looked up, and is completely irrelevant for your demographic. You’ll get ads for it anyway by the end of the day or week.
Listen to the Big Tech comments to press and congressional testimony very carefully. They always say something like, ‘Facebook is not spying on your microphone.’ They’re always very carefully wording it as the parent company is not listening to your devices. But they absolutely know either one of their subsidiaries or their partners are listening to your microphone, and feeding the data to them.
My roommate, an iPhone user and English only speaker, left his TV on the Spanish channel with his phone right next to it and left for work. When he came back home and checked his phone, all his Facebook and Twitter ads were in Spanish. This was at least 10 years ago.
I’m always torn on this topic because, yeah there’s hundreds of biases that can be attributed to this phenominon.
IE something becomes a popular topic in your area. Meaning more people in the area start searching the topic in that area, thus advertisers start pushing it to that location.
Obviously ads are also tracking you in 100 ways on what you’ve searched for, looked at etc… which means it could have a good guess of what you are going to talk about, before you do.
But at the same time, I think everyone can think of a lot of stories of things that just seemed to perfect, to out of the blue. For me the big one was 10 years ago when I walked into an attic, said “man it’s fucking dark up here”, opened my phone, and a big ad for a flashlight app popped up.
I think we will need a few more lawsuits such as Apple has agreed to pay $95 million to settle a lawsuit alleging that its virtual assistant, Siri, recorded users’ conversations without their consent before this is no longer treated as confirmation bias or people been paranoid.
My wife used to tell me that her adds would change after discussing something and at first I did not believe her, but it just kept happening again, and again. It reached the point that we would put our phones away, discuss something and there is no change in ads about the topic. If we had our phones near adds would change.This would happen on things that we would not see adds for normally. For example we would discuss a trip to a place we have never been and she would start seeing adds about the destination after that.
ADS
Ads is from advertisement, not adds
Yeah it would be pretty annoying if her active directory domain changed that often.
This article’s preamble cites this news story as the motivation for writing the rebuttal presented.
ARS said, that reuters said, that users said.
Someone needs a new hobby. “Proof” from 3 layers of journalists interpreting a case that they themself said never went to court. Trying to use evidence of absence as proof will never win any hearts in a debate.
I didn’t seriously believe it happened either for quite some time because confirmation bias is a bitch. But I’ve seen it happen a few times where it would have to be a seriously unlikely coincidence.
If it was searched for in Google, Facebook, apple, or whatever sure
If it was correlated with locality and time, sure.
You can infer a lot from a few searches but there are times where nothing was searched for and a novel concept came out of conversation and book there’s ads and search completion for it.
Maybe, just maybe, someone settling a lawsuit without being found guilty, doesn’t ACTUALLY mean they’re innocent.
Quick experiment. If you don’t own a dog or a cat, talk about buying dog or cat food a couple times today.
I have both, also, if it’s real, you’d have to match up with an advertiser that really wants your profile.
I search for crap all the time but don’t get ads most of the time, then one time, I look up this one kaz air filter and get nothing but ads for it for a week. hundreds of home depot ads.
I love how having both a dog and a cat is somehow dismissive
My big problem isn’t with the concept I could talk about buying parrot food.
But there has to be a vendor out there that says hey whoever I’m buying this data from, I need to put an ad in front of parrot owners.
These are going to be very high cost ads, so whatever products they’re going to sell you probably have a respectable profit margin or respectable expected lifetime value.
Trying to trigger it on purpose, without any idea of who’s advertising or for what is somewhat of a fool’s errand.
How many dog and cat owners are out there anyway? Dozens perhaps.
So Apple and Google have created the most sophisticated spyware known to man, so undetectable that tens of thousands of developers and researchers have never even seen a sign of it, and then they use the data for ads so sloppily that anyone can prove they’re listening?
I’m sure Apple just paid 95 mil because they were bored.
That Siri was bugged in a way that activated it unintentionally, which then sends recordings to Apple, is not in dispute. Turning that into “they’re always recording your conversations” is a big leap. Why would the whistleblower that revealed the recordings being misused not bother mentioning that?
It was activated at times in which it was unintentional and then they sold that data.
People are saying, and I have observed, extreme coincidences with the ads were timely, They were on novel data that wasn’t thrown through searches, and they weren’t explainable by locality.
You don’t have to be recording 24x7 to get they observed outcome.
Do you have any proof they sold that data? I’d love to know why the plaintiffs settled out of court if they thought they could prove Apple is feeding every voice recording into their ads. They had to pay 5x as much just for slowing down old iPhones, actively selling voice recordings would undoubtedly be worth far more than that.
The issue is that contractors had access to the recordings, which is certainly a breach of privacy, but not a grand conspiracy to target ads.
I’m not going to play move the goal posts with you all day long.
At what point did I move the goalposts? I never denied that the recordings existed. I simply fail to see how someone at Apple would decide that selling private conversations is worth the insane risk.
<img alt="Hmmmm." src="https://lemmy.dbzer0.com/pictrs/image/7c45d678-386a-4c47-b701-cf00b1a5bdf0.webp">
Literally the news story that this author cites as motivation for writing this article in the preamble to the article.
People always talk about getting served ads after they talk about something. I think it’s the other way around. The ads put the thought into your brain and then you start talking about it and notice after you’ve already been thinking about it for a while.
While I do suspect they listen, I have pretty solid (anecdotal) evidence they scan text messages. When I bought my house I had no solicitor, I text my buddy to see who he used and he texted me a response.
Started to type into Google to get a number and it was the top suggested search after 2 chars. Nowhere else did I mention this solicitor, hadn’t heard of them before this, have no other searches for this solicitor. It’s not a big firm, it’s not even in my city - only explanation I have is they scanned the messages.
Forgive me if this is here already but this is how your post showed to me.
<img alt="" src="https://aussie.zone/pictrs/image/9db59c7a-f2d2-4f9f-bb9a-737bc028d35c.jpeg">
I found indisputable proof of this happening.
We were using Google maps, driving in a production van. We were talking about the song “Gasolina” by daddy yankee. The person whose phone it was did not speak Spanish. Moments later we were being served suggestions to stop at “estaciones de gasolina”
That’s not how proof works.
It’s well possible and previously tv mic had been used as bugging device. The problem is, way too many security researchers look in system level software of iOS and even other components of the device that such practice will be too risky for apple (same applies for mainstream android products). Also processing realtime audio, finding potentially unrealiable topic from it and doing realtime ad is actually too much work as of today’s tech (might change sooner than you think though).
What, I think, is more practical is to use the whole query after the wake word to show ad, and potentially use other app tracking data, which is way much reliable than voice for targeting purpose. Voice data is useful for bugging purpose, primarily (ab)used by nation states and LE.
I bet in the medical procedure case mentioned in the blog the user searched/talked about that in other apps and average people aren’t good to notice these privacy leaks.
All the assistants listen all the time for their codeword. The new pixel phone show you a list of songs played around them and more. It is already happening all the time in the background.
That’s done locally. You can try training wake word models for any open assistant and see how much computing power it needs for even simple phrase.
I’ve always theorized that it should be possible to have multiple wake words with different functions, some invisible to the user.
It has to be “always listening” for the wake word to function at all, so it clearly is doing that, what’s to stop them from having another wake word like “bomb” which it then starts recording and sends to the NSA for instance, or even “clip the last 30 seconds” like an xbox could be feasible. Or even have corporations pay to get on the “list” of secret trigger words, like Toyota pays and it hears “Toyota” or “new car” and starts serving ads for 2026 Celicas (I wish lol). It doesn’t even have to send much data back for that, just “ohp, said word, check box to join ‘toyota’ ad group.”
I’m not saying they do that, but like, it sounds totally easily possible and I can’t be the first person that had this idea, why wouldn’t they?
Don’t give them idea :)
Yes that’s indeed a possibility.
The worse part is, they don’t really need to bug your mic to figure out what you are talking about to target ads to you. The best sales leads are the family and friends of your existing customers. So say you talk to you coworker about how they switched to this new diaper rash cream for their baby. You might not have a baby but you talked about it and somehow you got ads for diaper rash cream. What really happened though is that your coworker bought their cream on Amazon and that brand purchased target ads for everyone whose location data was nearby them. Or they bought it for everyone whose phone was connected to the same IP address. We have so much data tracked about us that they can guess what we are talking about without actually having to tap our phone lines
In addition to location, the data collected moreso resemble demographics than specifics. And on some of the most mundane shit at first glance, but actually gives a very clear picture of the consumer. Things like 1. OS installed 2. version of OS installed 3. Battery percentage 4. Total device memory 5. Remaining total memory and more things like that.
I liken it to how a psychic fools people into thinking they are magical when really they are incredibly perceptive and experienced in making judgements based on client’s clothes, appearance, demeanor, etc before they even open their mouths.
Using your search data is bad enough
I’ve literally seen advertisements for products that I was talking about but explicitly did not search for or type or anything on any device. All I did was talk about it in real life.
It’s literally a thing that happens, I have seen it happen first-hand.
“I’ve seen it first-hand” isn’t significant evidence because the frequency illusion effect is a thing. If you see dozens of ads a day and ignore them unless you notice them matching something you talked about, you’ll end up thinking ads can track what you talk about whether or not it’s true.
While i understand and agree with the premise, i think it’s lacking context. It is quite disturbing to have an obscure conversation (you know, we’ve never been to tahiti), and suddenly you’re getting banner ads or sponsored results about trips to tahiti.
This is absolutely a thing that happens. It happens to my wife frequently (the amount of times i hear giggling, i was just talking about that! Now I’ve got an ad! What a coincidence!), but i disabled all my google permissions (outside of location for maps), so it doesn’t seem to happen to me at all.
I don’t think every company does this, but some do. I also had to uninstall WhatsApp because my microphone usage was up while i was sleeping. That was quite concerning to discover. Whatsapp claims it’s a bug, but I’m not sure about that.
news.ycombinator.com/item?id=40348711
ghacks.net/…/report-alleges-that-microphones-on-d…
I would agree with you about the frequency illusion effect IF it weren’t something very specific and niche.
It is literally a thing that happens.
I have worked for an advertising company before (they hid that they were an advertising company) and you would be surprised how sophisticated and scummy ads can be.
.
The speech recognition software used by digital assistants that come with most modern smartphones would make it trivial to process the audio locally and map the output to your ad profile. Much lighter lift than sending audio recordings.
And a much smaller footprint. It could even be binary data for tweaking your algorithmic profile, say the name of a branded product or in the case of a product with few options just the type of item. Audio runs in the megabytes per minute, transcripts in the kilobytes, but reducing to a conclusion of interest in a single specific item is really very small, hard to notice tbh.
In September, I was using reddit, had an iPhone, etc. I was generally aware of digital privacy, probably moreso than the average person, but by no means was I knowledgeable.
I was running a beta on my iPhone at the time, for context. I had a short conversation with my roommate while my phone was in my pocket. I took it out to text my partner and pressed the dictation button. My phone proceeded to type out the majority of the conversation I had had maybe five minutes earlier with my roommate. Literally ruined my ignorance is bliss and now I have a Pixel with grapheneos and use almost exclusively open source software with a major focus on privacy. Obviously this is an anecdote from some idiot online and I can’t verify what I’m saying at all, but the experience definitely shook me.
This is a great case of confirmation bias, too. The one time your ad happens to match a conversation you had earlier, you’ll be convinced forever, and tell everyone you know about it. The ten million other times you have a conversation that doesn’t appear in your ads will go unnoticed.
Can I ask, why are you so ready to performatively forgive them here? Apple is not your friend, Apple and Tim lined up to donate the million like the rest of those greedy, transactional cowards.
Apple doesn’t “do” it per se, instead Apple shares certain data with third party partners for the purposes of “improving your product experience” the data is then laundered 17 times through middle layers and added to a shared digital fingerprint of you and your household’s web of connected devices. You and your family are then sold on a marketplace as advertising targets actively interested in X category or product (Apple is also subsequently a customer in that marketplace). You then either receive that advertising or your family is targeted with it so that they can then casually mention the product back to you (company knowing you were already interested) so it feels organic and “I was just thinking the same thing!” and boom, you’re buying that new set of pots and pans.
We’re already living in the matrix, you’re just a little drone being pinged around according to other people’s will, to support the pursuit of endless growth. So yes, in a way companies are spying on you… After you’ve given them individual permissions to access your microphone and permission to share “certain data” about you with third parties, in a carefully orchestrated dance - so that they have plausible deniability and so you don’t have to threaten your parasocial relationship with their brand and can continue saying “probably not Apple though…”