SoylentNews is people
SoylentNews
https://www.bbc.com/news/technology-49585682
A recent study has "proven" that Android and Apple phones do not eavesdrop on conversations.
A mobile security company has carried out a research investigation to address the popular conspiracy theory that tech giants are listening to conversations.
The internet is awash with posts and videos on social media where people claim to have proof that the likes of Facebook and Google are spying on users in order to serve hyper-targeted adverts. Videos have gone viral in recent months showing people talking about products and then ads for those exact items appear online.
Now, cyber security-specialists at Wandera have emulated the online experiments and found no evidence that phones or apps were secretly listening.
I think this is the classic can't prove a counterfactual thing, ("we've proved that no birds swim from this survey of life in a nearby park"). However, I thought this would be an interesting article, and possibly start some good conversation... if nothing else, due to poking potential holes in their experimental techniques and data analysis:
Researchers put two phones - one Samsung Android phone and one Apple iPhone - into a "audio room". For 30 minutes they played the sound of cat and dog food adverts on loop. They also put two identical phones in a silent room.
The security specialists kept apps open for Facebook, Instagram, Chrome, SnapChat, YouTube, and Amazon with full permissions granted to each platform.
They then looked for ads related to pet food on each platform and webpage they subsequently visited. They also analysed the battery usage and data consumption on the phones during the test phase.
They repeated the experiment at the same time for three days, and noted no relevant pet food adverts on the "audio room" phones and no significant spike in data or battery usage.
The activity seen on phones in the "audio room" and the silent rooms were similar. They did record data being transferred from the devices - but it was at low levels and nowhere near the quantity seen when virtual assistants like Siri or Hey Google are active.
James Mack, systems engineer at Wandera, said: "We observed that the data from our tests is much lower than the virtual assistant data over the 30-minute time period, which suggests that the constant recording of conversations and uploading to the cloud is not happening on any of these tested apps.
(Score: 0) by Anonymous Coward on Friday September 06 2019, @02:17PM (9 children)
Check youtube.com for proof otherwise.
(Score: 3, Insightful) by ikanreed on Friday September 06 2019, @02:46PM (8 children)
Citing youtube as a source for anything, even if you can direct people to a specific video, is a sign of extremely deficient critical thinking.
The only exception I'm willing to consider is as a primary source that someone said something, and even that's prone to deceptive editing. Any youtube analysis, punditry, rant, recording of an event, feat of skill, or presentation of fact should be assumed to be completely and utterly full of shit until proven beyond a reasonable doubt otherwise.
(Score: 0) by Anonymous Coward on Friday September 06 2019, @02:53PM (7 children)
Primary Source: https://www.youtube.com/watch?v=XQcGCnvF264 [youtube.com]
(Score: 3, Insightful) by ikanreed on Friday September 06 2019, @03:36PM (6 children)
1. That's not what a Primary source [wikipedia.org] is.
2. I think the rest of the content that channel produces [youtube.com] more or less proves my point about youtube videos being inherently untrustworthy medium.
(Score: 0) by Anonymous Coward on Friday September 06 2019, @04:00PM (5 children)
What is untrustworthy about it?
(Score: 2) by ikanreed on Friday September 06 2019, @04:16PM (3 children)
The fact that they're click bait hell videos? The fact that they claim to do clearly technically impossible things?
(Score: 0) by Anonymous Coward on Friday September 06 2019, @07:25PM (2 children)
So, basically you are racist against indians.
(Score: 2) by ikanreed on Friday September 06 2019, @08:14PM (1 child)
I think I've made this incredibly clear, I'm racist against youtubers
(Score: 0) by Anonymous Coward on Saturday September 07 2019, @04:09AM
Yes, as I said: you are racist against Indians. All the top youtube channels are Indian: https://www.socialbakers.com/statistics/youtube/channels/ [socialbakers.com]
(Score: 0) by Anonymous Coward on Friday September 06 2019, @05:34PM
The amount of yellow and moonspeak are a start. And all of the english words that show up are pure clickbait. I don't know what the fuck that channel is about, but its about getting clicks and not giving any useful information.
(Score: 4, Insightful) by Runaway1956 on Friday September 06 2019, @02:38PM (6 children)
The methodology seems strange. If I'm eavesdropping on someone, I'm doing my very best to IGNORE the sounds of pets in the background. While nosy bastards are training AI to do a lot of potentially intrusive things, I've not heard of an AI trained to identify how many, and what kind of pets you might have. This little "research" item wasn't designed to test for eavesdropping. It was designed to figure out if artificial intelligence would cue on the sound pets.
(Score: 3, Insightful) by seeprime on Friday September 06 2019, @03:18PM (1 child)
It really was a terrible study that seems to, possibly intentionally, ignore that algorithms listen for human words to decide what ads to display in response to the input. Animal sounds will be ignored.
(Score: 5, Insightful) by EvilSS on Friday September 06 2019, @03:41PM
(Score: 2) by EvilSS on Friday September 06 2019, @03:42PM
(Score: 2) by HiThere on Friday September 06 2019, @09:14PM (1 child)
They measured battery use also, so it is decent evidence (not proof) that the basic systems don't constantly eavesdrop.
This say nothing about systems with various apps installed, of course, and many of those apps are not removable.
Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.
(Score: 2) by HiThere on Friday September 06 2019, @09:17PM
I REALLY wish it were possible to edit things after a post.
I wrote the post, wasn't satisfied, so I went back and edited it. But now the text says "This say", where it should say "This says". Fixing one thing borked another.
Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.
(Score: 2) by driverless on Saturday September 07 2019, @03:24AM
Yep, definitely suspect methodology. It's like putting a child molester in a room with surveillance cameras recording everything they do and then sending in a kid, and concluding afterwards that because they never touched the kid they're perfectly safe around children. Of course nothing's going to happen in a carefully-controlled environment, it'll happen when you're not watching closely.
Having said that, it does have some value in that it debunks the most obvious conspiracy-theory stuff. We know companies like Google use our browsing and messaging for advertising, so it's not like they're not doing it at all, but obviously it's not as blatant with voice stuff. I wasn't even aware of this particular conspiracy theory until I saw this story, you filter data that's passing through your systems anyway because there's a ton of it readily available, but don't proactively go out and collect it by turning phones into bugs.
(Score: 5, Informative) by VanessaE on Friday September 06 2019, @02:39PM (3 children)
So don't look at it that way.
Likely the real intent was not to prove an absence. Rather, to seek to DISprove the conspiracy theories themselves, and they used what seems like a good overall methodology, on the surface anyway. I do see two issues though:
For one thing, I would think marketers' audio analysis tech that would potentially be listening-in would already be able to recognize and ignore what sounds like an advertisement -- if Google search and Youtube's copyright guard can recognize and name a song, recognizing and rejecting input from an ad shouldn't be a whole lot different. As such, one would think that any ad-targeting algorithm that were in place would be getting almost nothing useful. The test audio should have been a big chunk of actual pet-related conversations/discussions.
Second, the sample size is just too small. Two pairs of phones put to the test over the course of three days is a bit thin. Needs more testing, perhaps at least a month's worth of listening, with much longer audio, and with at least a few more pairs of phones placed around the room at different distances relative to the audio source (with one pair placed at a known distance, as a control). They should also do this same type of test with other subjects, for example, another set of phones listening in on, let's say, baby-care related chatter, another set snooping on something gaming-related, and yet another set listening to something car-repair-related.
(No, I didn't RTFA, but does anyone? 😛 )
(Score: 4, Insightful) by JoeMerchant on Friday September 06 2019, @05:45PM (1 child)
Rather, to generate a reassuring headline to tip public opinion slightly in favor of whatever agenda is really behind the press release...
🌻🌻 [google.com]
(Score: 2) by jmichaelhudsondotnet on Saturday September 07 2019, @08:37AM
yes
The several other informal studies that have propagated where people noticed their phones were working too hard, must have reached the level that a comforting counter message must be airdropped. This indicates perhaps actually a good trend, that people are distrusting their phones, which they should be.
So they wouldn't be working so hard on their manipulative lies if the truth wasn't making headway somewhere, so maybe this is the best news i've heard all day.
(Score: 2) by jmichaelhudsondotnet on Saturday September 07 2019, @08:34AM
Indeed, this seems designed with a very generalized, very sanitized study, to give the maximum impression that people can control their phones. Don't worry, this isn't the vulnerability you are looking for....
It's noteworthy how the phones in the lab are treated like alien artifacts, we can't actually take them apart like other things to study them, they are precious and monolithic. How reverent the testers are of IP. This whole problem even exists because people like me *cannot* escalate our admin privileges of this miniature i386 form factor to look at its processors, transmitters, ram and hard disk. To me, the need for studies like this is the entire purpose these privileges are not available, otherwise all of the tracking would be out in the open and we would be designing open source tools to remove them that you did not need a heavily restricted and tracked 'app store' (aka corporate repository) to install, just like the rest of the i386 platform. But no, we can't have nice things, we have to trust the hand that feeds us if we want a pretty phone.
This I eventually declined and I recommend everyone do so as well. For me getting a 'smart' phone would be like getting a windows xp computer i had to attach to my hip. I write about it on my site, 'smart phones and wild bears'.
I agree with both of your two points but find other points more compelling.
3 the nso group and friends can send you a single text message that comprimises your phone completely, without any action on your part
4 the data profile of the fictional person holding the phone in the tests is empty, so it isn't interesting to anyone. the dragnet of surveillance will kick in and start setting off passive flags for marketers or spy agencies to target you once you start acting like a human being.
So for all of these points, I find the test this organization ran to be anti-scientific and awful, it is there to shut down intellectual inquiry while providing little to no information. The organization that did it is likely somehow connected to the people doing the tracking themselves, although the people doing the testing could be well meaning.
I might call this kind of science, like the 5g tests of a single transmitter under optimal conditions, 'don't worry your little head' science, inappropriate for humans but directed at us nonetheless.
(Score: 2) by takyon on Friday September 06 2019, @03:32PM
But the FBI/CIA/NSA is ready to offer their services if you say the "wake words".
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by hwertz on Friday September 06 2019, @04:57PM (3 children)
Yeah.. if someone is so obsessed about some commercial product that they are talking about it to people like some shitty TV ad, then I'm quite sure they will have done at least some of googling the product, liking the product on facebook, posting about it on facebook or instagram, or going to pages that discuss said product. All of these WILL be tracked by facebook as well as several ad networks.
Although this study is pretty poor, I think it's also quite unlikely any of these apps are abusing the mic, they can get all the info they want about you by tracking everything else.
(Score: 1) by nitehawk214 on Friday September 06 2019, @05:40PM (2 children)
On the rare times I see ads (usually when on mobile), they are nearly always for things I have already purchased. All google can extract from me during my browsing habits is the make of car I drive, the model of my favorite camera, and that I am obsessed with telescopes. 90% of ads I see are for these things, and usually specific brands I have.
It is somewhat annoying that I have received emails from stores I have purchased from in the past offering me deals on products that showed up in a google image search, but didn't get clicked on. (looking at you, Highpoint Scientific [highpointscientific.com]) I have since unsubscribed from their marketing emails.
"Don't you ever miss the days when you used to be nostalgic?" -Loiosh
(Score: 3, Interesting) by HiThere on Friday September 06 2019, @09:22PM (1 child)
OK. But consider that I get ads from companies where I have purchased products, not from other companies selling similar products. To me this indicates that the leak of information was from the company where I bought the stuff.
It seems quite foolish. I bought a pair of shoes, liked it, so I bought two more pair. Now I'm getting ads for shoes all the time from that company, when I don't want to buy shoes. (And if I already needed another pair of shoes, I'd definitely want a different brand.)
Javascript is what you use to allow unknown third parties to run software you have no idea about on your computer.
(Score: 1) by nitehawk214 on Monday September 09 2019, @01:43PM
I suspect this means that your browsing habits, use of adblock, use of script blocking cause Google to have less information to give to advertisers for you. Since they know you purchased a product from certain companies, you are getting ads for only those companies.
"Don't you ever miss the days when you used to be nostalgic?" -Loiosh
(Score: 3, Informative) by Knowledge Troll on Friday September 06 2019, @07:58PM (10 children)
In order for Google to not be using open microphones to tailor advertising to us they would have to literally not do what they said they would do. The how and the why is contained in a Google paper from 2006 called Mass personalization: social and interactive applications using sound-track identification [googleapis.com] by Michael Fink & Michele Covell & Shumeet Baluja.
What is described in the paper is slightly different than the technology that would be used to enable the kind of personalized advertising people are talking about in the context of this article but it isn't different enough where it seems impossible especially since this technology is over 10 years old now. As well Google publicly stated they were going to go forward with this strategy and technology in tandem with the release of the research paper.
Google already told us they would listen to every microphone and use it for advertising. Why should I believe they aren't doing that?
(Score: 2) by darkfeline on Saturday September 07 2019, @03:58AM (9 children)
So the fact that the military has documentation on what to do in the case of zombie attacks is proof that zombies exist right? Every possibility that is ever brought up in literature is clearly true and extant.
Hint: Just because it was proposed does not mean it was done.
Join the SDF Public Access UNIX System today!
(Score: 2) by Knowledge Troll on Saturday September 07 2019, @10:11AM (8 children)
I can't find the interviews with the Google CEO at the time that said they fully intend on bringing mass-personalization to market as soon as they can. But here you can have a twitter post from 2018 explaining how important mass-personsalization is to Google: https://twitter.com/thinkwithgoogle/status/980429746625531904?lang=en [twitter.com]
They said they have the technology in 2006, they said they would, and they continue to say this is important. Do you think Google would not do this if they thought they could get away with it?
(Score: 2) by darkfeline on Saturday September 07 2019, @09:32PM (7 children)
You're fallaciously equating mass personalization with passive eavesdropping.
Join the SDF Public Access UNIX System today!
(Score: 2) by Knowledge Troll on Saturday September 07 2019, @10:29PM (6 children)
Google is the one who equated mass personalization with passive eavesdropping by saying that they would use passive eavesdropping to implement it. Here I found the link to the interview with Peter Norvig, director of research at Google in 2006 talking about the technology I mentioned here [technologyreview.com].
I don't really see how much more Google could express their desire to engage in passive eavesdropping more than explaining that they are going to do it.
(Score: 2) by darkfeline on Saturday September 07 2019, @11:39PM (5 children)
I suggest you look up the definitions of the words "predict", "prototype", "early-stage", "research", "weren't", "concept", and generally actually read the words that you've quoted there.
Join the SDF Public Access UNIX System today!
(Score: 2) by Knowledge Troll on Sunday September 08 2019, @12:49PM (4 children)
You are the one glossing over words - I should have set some of them in bold perhaps. Did you bother to read the article as hyperlinked?
Ok
I see.
Oh. But it's only been 13 years. That isn't enough time to get this refined and out into the public. Nope definitely Google isn't doing this and there wasn't ever any intention stated to do so.
Your zombie analogy is failing because Google has plans for zombies, worked on zombie technology, and said they expect to use zombies.
(Score: 2) by darkfeline on Sunday September 08 2019, @08:58PM (3 children)
You conveniently omitted the "research director predicts...". Of course the research director would predict some application for research. There isn't any researcher who would brag that their research has no practical applications over the inverse.
In case you didn't know, research teams do not guide the development of products, merely do research which may then be incorporated into products--or not.
Join the SDF Public Access UNIX System today!
(Score: 2) by Knowledge Troll on Monday September 09 2019, @03:42AM (2 children)
What are you asserting? Presumably you aren't just arguing - there is a purpose here right? Are you asserting that Google is not engaging in this kind of surveillance? Are you asserting there is insufficient evidence?
My assertions were that Google has the technology to engage in the surveillance people are anecdotally reporting and that they stated they will use the technology. Turns out they stated they intend to use the technology - that's not an important enough difference for me.
My question was "why should I believe Google isn't doing this?" - the answer seems to be "because Google might not use technology they invested in 13 years ago regardless of how amazingly lucrative the idea is or how much they demonstrated they don't have issues with surveillance."
(Score: 2) by darkfeline on Monday September 09 2019, @07:40AM (1 child)
> they stated they will use the technology. Turns out they stated they intend to use the technology
No they didn't. A research director stated that they predicted that an early stage research concept may have applications, and the statement was not made on behalf of the company as a whole.
> What are you asserting?
I'm challenging your unsubstantiated claim with the intent to incite FUD and/or emotional reaction. I believe the contemporary vernacular is "fake news".
You stated in your initial post:
> Google already told us they would listen to every microphone and use it for advertising.
And the evidence you mustered to back this claim is one researcher predicting the application of an early-stage research concept.
While this kind of bullshit is now common in most news media, I will challenge it here on SN. If you have actual evidence to back up your claim, I'd like to see it.
Join the SDF Public Access UNIX System today!
(Score: 2) by Knowledge Troll on Monday September 09 2019, @02:02PM
Oh that's why you are being such a fucking asshole - you feel personally vindicated because you believe someone else is doing something you don't agree with by jumping to conclusions. Ah ha.
If you would like the larger story try reading this journal entry: Hypothesis: Google is using ambient audio for targetting ads [soylentnews.org]
Fuck you very much too.
(Score: 0) by Anonymous Coward on Saturday September 07 2019, @12:57AM
LIES!
(Score: 2, Insightful) by anubi on Saturday September 07 2019, @04:35AM (1 child)
One app I would sure like to have on my Android would intercept my internet connection and tell me what apps are connecting, who they are connecting to, and how much data is being sent or received...kinda like Wireshark.
"Prove all things; hold fast that which is good." [KJV: I Thessalonians 5:21]
(Score: 3, Informative) by jmichaelhudsondotnet on Saturday September 07 2019, @08:19AM
You notice when you use wireshark you have to authenticate as a root user, because you are digging down deeper into the internals of the system.
This ability to escalate privileges is not given to you as the owner of the phone by these companies, this is why such a backwards, old fashioned test is being used.
You should notice how the phones themselves are treated in this like a mystery, black box, 'how could we ever understand what is inside' kindof way, the testers accept the reasoning that they can't access the internals of this thing they own.
I would assume personally that any test done on phones that are in some kind of lab with only the most popular apps, just doesn't get anywhere near real world scenario. Like testing a car when its wheels can't touch the ground.