(I know many of you already know it but this incident I experienced made me so paranoid about using smartphones)
To start off, I'm not that deep into privacy rabbit hole but I do as much I can possibly to be private on my phone. But for the rest of phones in my family, I generally don't care because they are not tech savvy and pushing them towards privacy would make their lives hard.
So, the other day I pirated a movie for my family and since it was on Netflix, it was a direct rip with full HD. I was explaining to my family how this looks so good as this is an direct rip off from the Netflix platform, and not a recording of a screening in a cinema hall(camrip). It was a small 2min discussion in my native language with only English words used are record, piracy and Netflix.
Later I walk off and open YouTube, and I see a 2 recommendations pop-up on my homepage, "How to record Netflix shows" & "Why can't you screen record Netflix".
THE WHAT NOW. I felt insanely insecure as I was sure never in my life I looked this shit up and it was purely based on those words I just spoke 5min back.
I am pretty secure on my device afaik and pretty sure all the listening happened on other devices in my family. Later that day, I went and saw which all apps had microphone access, moved most of them to Ask everytime and disabled Google app which literally has all the permissions enabled.
Overall a scary and saddening experience as this might be happening to almost everyone and made me feel it the journey I took to privacy-focused, all worth it.
Yup. I was driving in the car with a few people for work. We were talking about a music video a couple of us had worked on, and we were explaining who daddy yankee/bad bunny was, and we mentioned daddy yankee did the song “gasolina.”
We live in the US, the conversation was in English, but fuck if “estacion de gasolina” didn’t show up on our route.
It's possible that it's inferred off the digital footprint of you pirating the content, also. People freak out a lot about being listened to, but I'd argue that's an inefficient spying mechanism they probably don't lean heavily on if they can avoid it. We're all living on platforms that are knowably spying on everything you click on or read or do online and feeding that into giant AI models with everything about you. Like just by watching a pirated video on a Google TV device, Google's hashing that and phoning that data home, possibly even matching that to the specific file, and adding that to an ad profile.
Yet again, someone mistakes an anecdote for evidence. And evidence is also not the plural form of anecdote.
I'm sure we have people here who are tech-savvy enough to have actually examined the kinds of data that their phone is sharing.
If you have something like Google Home or Amazon Alexa, then yeah, those would be sending voice data back, and yeah, they could probably use it for advertising. But as far as I know, there is no evidence that phones are "always listening" and "always sending information back" when they're idle.
And how often. have you said stuff that you have not received advertising for? You will notice it when you get a positive match but not on a negative.
Data collecting companies can predict/rate your behavior for more then 20 years based. Since then. it has been perfected. They know that you are interested in those topics without having the need to waste resources on recording and analyzing every single audio stream.
Listening to audio would be the least effective and most expensive method of data collection for advertisers. It's not happening. They already have literally over a million data points on you, there's nothing useful for them to glean from your audio that they don't already have ad nauseum.
You see thousands of ads and recommendations every day. You finally found one that was relevant to you. It's not that deep.
I’ve gotten ads for things I’ve just thought about. Never said anything out loud about or did any searches related to. It was something in a video I’ve watched dozens of videos about in the past. But on this occasion, I happened to think that I kind of want one for the first time. And I just so happened to start getting ads for them right after, also for the first time. They know way more about you than you think and don’t need to listen to you.
No, your phone doesn't listen to you 24/7. With that out of the way, there are a number of places where youtube may have gotten that info. One possibility is that someone in your household looked up the movie and maybe checked if stuff ripped from netflix is indeed full HD. And since everyone in your family is using the same NAT IP, then it's easy for youtube to target recommendations at everyone in that household.
I don’t doubt you, but it’s worth asking if your reasons for stating that our phones don’t listen to us 24/7 haven’t changed since you first formed the opinion.
Lots of things are meso-facts (a true fact at rhetorical time we learn it, but no longer true later). Tech moves quickly. It’s worth not assuming anyone is right here, & asking: under what conditions could our phones be listening (enough to produce what OP experienced)?
The conditions would be that all the controls that are in place to prevent it from happening are bypassed, which no one has proven yet. For example, Apple has developed their devices (assuming not jailbroken) in such a way where the camera and microphone usage indicators are hardwired and can't easily be bypassed by software hacks. So if your phone was listening to you all the time, then the microphone indicator light would always be on. Listening 24/7 would also drain the phone's battery and use up so much data it would be noticeable. Another example is Siri. It is actually designed in a way where there are 2 components. The first one is local on the phone and separate from the actual Siri component. It is what's actively listening for you to call it. Once you call it, it then activates the actual Siri that transmits your voice inputs online.
Here’s a fun little experiment you can try. Make a list of random topics and have a discussion about each of them on separate days. Make sure each topic is something that could result in creepy suggestions or ads on YT. If even one of these topics produces the expected result, you could be on to something.
Fun, sure, but not an experiment that would actually be meaningful.
The data from your phone's microphone doesn't magically appear in Google's advertising servers. It would have to go through a lot of steps before it gets there, and one of the first steps is in your home (if you're on WiFi). One can analyze the traffic/data that leaves their phone.
It's good to be cautious, but worrying about your phone's microphone is potentially like worrying about your windows while leaving your front door open.
There is one more thing I haven't mentioned here. The device where I pirated the movie is different and is on different Google account and my Google account on which I opened the YouTube was different.
You just mentioned 2 different Google accounts: if your devices are connected to Google accounts they are already getting a lot of information from you that way, and Google knows that those 2 accounts are related.
I will watch these later. But recently one of the Facebook's employee's chat was leaked saying they listen to customer mics 24/7 via a third party. Google blocked the alleged third party and Facebook has ended ties with them too.
It was an ad partner's pitch deck, not much to do with Facebook itself. And it didn't really explain how it would be listening anyway.
Besides, if they were recording, processing and / or transferring audio, that would mean there's data usage, battery usage, etc - stuff that's easy to prove.
The truth is a lot simpler (and scarier) and you will find that in the links I provided.
The youtube algorithm determined the following: people who watch the kind of videos in your history, are also interested in recording netflix shows. And it was right, because you are in fact interested in that (general) topic. This is another possible explanation.
This may be a simple coincidence. Maybe you had similar YouTube suggestions in the past but you didn't pay attention because they come at random times. Like if you drive a Honda Civic, you tend to spot all the Honda Civic in the street.
There would be an interesting experiment to make though:
Take a snapshot of your YouTube recommandations
Choose a subject that has nothing to do with any of the recommandations, let's say "travel to the Bahamas"
Hold a conversation with someone with both your phone's present, mention several time going to the Bahamas.
Check YouTube again, si if the topic of Bahamas is appearing.
Choose another topic not covered by your recommendations, let's say collecting stamps
Put your phones away, have a conversation about collecting stamps
The problem comes is the suggestion of travelling to destination X (in your case, the Bahamas) doesn't just pop up out of thin air - friends may have travelled there recently, perhaps there has been a recent advertising push, etc.
Another family member looking up some destinations to travel, then speaking with you later - same external IP of the home wifi being reported, bam you get advertised the destinations they looked at the most.
Choosing a "random" topic again also doesn't come out of thin air.
First off, if you're concerned about phone privacy, consider a custom OS for your phone that respects privacy such as GrapheneOS.
It's easy to figure out that your device isn't listening to a constant audio stream 24/7, since that would drain battery and send a lot of noticeable data over the network. However, it is entirely possible to listen for certain keywords as you mentioned, and send them encrypted with another seemingly legitimate packet. There's no way to be 100% certain, but it is possible in theory without draining too much battery.
The steps you took are good, making sure that apps don't have any permissions they don't need. Privacy is a spectrum, so it's not "all or nothing". As I mentioned before, if you're seriously concerned about mobile privacy and want a solution, you can get a custom operating system that can remove any privacy invasive elements. GrapheneOS also allows you to disable the camera and microphone system-wide (although this functionality is present on some other Android builds).
If it eases you any, a lot of these advertisements happen to be coincidence and trigger confirmation bias. It could be that those ads happened to show up by coincidence, or that advertisers managed predicted your interests, or that you got tracked by some other means while downloading the movie. The possibilities are nearly endless.
You should install Rethink and see how much garbage your phone constantly transmits and receives. And this is not even a kernel-level firewall, so who knows how much data Google actually exfiltrates...
I don't know about a constant audio stream, nor about keywords, but I noticed that Google Keyboard sends out some data every time you type anything. It's not even that subtle.
If anything, I love GrapheneOS for its "Network" permission toggle. It's nice knowing that my keyboard (or any other unnecessary apps) can't phone home.
I'll second the recommendation for GrapheneOS. One of the available options I use is to keep mic, camera, and location off at all times until I need them.
That simple toggle ability changes your privacy stance greatly.
Don't get things for free! Ever! The producers aren't rich enough yet to pay the artists a living wage for their creative work. Homeland will extra-judicially use weapon systems on you even if you don't pirate because all it takes is false accusation and then you will be tortured and never allowed to reproduce anywhere in the US sphere of influence (or TPP or UN).
I understand completely. It's happened to me too many times now to be a coincidence. For a while there whenever I talked to myself "alone" I would sometimes pretend I was also talking to LaMDA, the AI. Now I know it's probably not LaMDA. But it's not outside the scope of possibility for Google or Meta to be listening 24/7 if it's an AI overseeing the panopticon.
I said it has happened to me mulitple times, and I have done experiments. I have given the listeners false information to see if and when it will spread. I have seen whole products pop up out of nowhere in nearby stores because of this. I have seen shitty half-assed tv shows and movies suddenly show up on billboards, as if with only an understanding of the bullet points of my overall speech, and even then with the shaky understanding of an AI trying to fill in the blanks.
Only thing I can think to suggest (that hasn't been suggested already) is to purposely poison the well by sometimes loading it with falsities for obscuration.