Hacker News Reader

Comment by MisterTea | Hacker News Reader

Good.

> She said the adverts she got "suddenly started changing within weeks to lots of baby photos and other things - ads about babies and pregnancy and motherhood".

> "I just found it unnerving - this was before I'd even told people in my private life, and yet Facebook had already determined that I was pregnant," she continued.

Facebook likely knew she was pregnant by covertly eavesdropping on her life via her phones microphone. When these companies started using their apps to eavesdrop on your entire life via your phone, I immediately deleted them. No way is that creeptastic shit allowed in my life. I mean, imagine talking about doritos in a nostalgic context only to have amazon suggest you buy them when you open the app. Deleted. The only apps on my phone are banking and utility. Bye Felicia.

View on HN

> Facebook likely knew she was pregnant by covertly eavesdropping on her life via her phones microphone

That's quite an assumption to make. The very famous example of data tracking and creepy correlations is the story about the supermarket sending baby ads to a household based on nothing more than their purchasing habits.

People change the way they act in measurable ways for much less intense psychological events than pregnancy. Just the grocery prices you look up (and that doesn't include things like pregnancy tests) is enough to determine likelihood of pregnancy, let alone an app you take with you all day and possibly spend hours scrolling through. The amount of data you're passively giving off to these apps is enormous.

I have yet to see any evidence of spying through microphones by large companies. I know some small ad network got caught doing so, but not the big suspects everyone is (rightfully) afraid of. The Google phone I'm carrying would've spawned much more relevant ads than "buy a new €200k electric car" and "where to apply for welfare" right after each other if it actually listened in.

>That's quite an assumption to make. The very famous example of data tracking and creepy correlations is the story about the supermarket sending baby ads to a household based on nothing more than their purchasing habits.

While a fun anecdote, it really shouldn't be taken seriously as something that actually happened.

https://medium.com/@colin.fraser/target-didnt-figure-out-a-t...

Did you actually read that Medium post? It's idle FUD and speculation that the author of the original 2012 Forbes article lied about the Target incident, without anything to substantiate that claim. It's also evidently written by someone working in data science at Meta, who would naturally have incentive to deny the reality of their industry.

Most of the Medium comments are calling them out on this:

>Title: "Target didn’t figure out a teenager was pregnant before her father did, and that one article that said they did was silly and bad." A clear assertion. A solid claim. One of the first paragraphs: "I won’t belabor this since we can’t actually know, but the anecdote about the father calling the store and talking to the manager is probably just not true. " So, a person who contradicts himself right in the beginning uses articles that are being hyperbolic as proof that the articles got it all wrong. Smells a bit like double standards, no? And the point of the conversation is less about accuracy of the findings and more about such models being in place. I mean, Facebook did this, right? https://www.snopes.com/fact-check/facebook-categorize-users-... And I'm not sure but there was this little company, very obscure, not sure anyone heard of it. Cambridge Analytica? Again, very obscure.

The target one always seemed like bullshit to me. They don’t have any magic algorithms. There were likely some hard tells that she was pregnant. As in something like a doctor’s office told them.

Or they just sent the same mailer to anyone with a female name popular in the last 40 years.

After all, at any given time, like 11% of women are either pregnant, will soon become or have recently been pregnant.

As someone who recently had a kid, I guarantee you’ll get flooded with very targeted baby stuff quickly. Not just fliers, but literal product samples.

I assume insurance or something sells the data.

I worked in computer infra stuff at Target at the time.

For one, they were incredibly paranoid about customer data privacy at the time, because of the famous incident of an unknown entity siphoning off data from some of their in store payment terminals for an unknown length of time. They instituted barriers and strict need to know basis access to everything touching customer history.

The amount of internal systems and departments that would have had to cooperate in order to start trying to sell a customer pregnancy related products because another arm of the company detected an individual customer was pregnant just boggles my mind, it could not and would happen at Target.

>Did you actually read that Medium post? It's idle FUD and speculation that the author of the original 2012 Forbes article lied about the Target incident, without anything to substantiate that claim. It's also evidently written by someone working in data science at Meta, who would naturally have incentive to deny the reality of their industry.

Did you? From the article:

>[...]The Forbes writer who rehashes the anecdote even describes it as “so good that it sounds made up.” Indeed.

>But this is an easy criticism and beside the point, so for the remainder of this article, let’s suppose that the anecdote is true.

The bulk of the objections in the article are that a single instance of Target correctly predicting a pregnancy isn't enough evidence to conclude they have some sort of a galaxy-brained AI that can predict pregnancies before the mothers even know. It's far more plausible it randomly guessed someone was pregnant, and got lucky.

The comment you quoted has the same issue. It gets way too hung up on the first part, ignores everything else the author wrote, and then tries to move the goalposts to "the conversation is [...] more about such models being in place". That's a huge cope. It's fairly obvious how concerned we should be about such models is proportional to how accurate they are, and the author is trying to harness that with the Target anecdote. Nobody would care if the story was "24 year old man got sent ads for Zyn, without telling Target that he smoked".

I have heard numerous anecdotes from multiple people I know of them being served targeted ads shortly after saying something similar aloud with a device microphone nearby. You're right that there are a lot more ways than a microphone for ad targeting to work, but I wouldn't necessarily dismiss that guess as wrong.

Why not dismiss it? Like you say, the targeting could work in legal ways. That would be feasible and sensible. But what makes you doubt it is that you listened to a bunch of dweebs with superstitions?

I don't think it makes a difference.

Spying through your microphone ala 1960's FBI wiretap is not worse than reading credit card history, location data, google searches, etc. Sure, it sounds more spooky, but probably they would have less data if they only listened through the microphone. That's probably an improvement over our current situation.

For what it's worth, I don't think there's any hard evidence that big companies are doing this. Aside from the technical difficulties of constantly listening to 1+ B mobile devices and running speech-to-text, it'd also be a legal quagmire.

What's more likely happening is far more boring: humans as a whole are very predictable, and the metadata that Facebook et al have access to is more than enough to identify these patterns.

To _really_ try and test this, you could download the Facebook (or Amazon or whatever) app, and (within earshot of your phone) consistently talk about some completely random purchaseable item that you have no intent to purchase and have never otherwise indicated you have an interest in. Chances are you will not get ads for it.

To make it less likely you pick something that is subconsciously on your mind (and thus able to be picked up via your metadata), try playing word association games to pick the product in question. E.g. banana > yellow > raincoat > weather > digital thermometer

> To _really_ try and test this, you could download the Facebook (or Amazon or whatever) app, and (within earshot of your phone) consistently talk about some completely random purchaseable item that you have no intent to purchase and have never otherwise indicated you have an interest in. Chances are you will not get ads for it.

That's the trick isn't it. How do you choose a random purchasable item without indicating interest in.

Is it because it was something you saw recently, but don't want to buy? How many (targetting) similar people near you also saw that and may want to buy?

How much is that ad being shown to you now, but you hadn't noticed because you weren't interested in the product?

It's certainly possible to carefully craft an experiment, but it's not easy. And a poorly crafted expirement won't show much.

> via her phones microphone

Source? Because so far this claim has been debunked every time.

I assume you are aware how Android and iOS have a visual indicator to show that the microphone is active? Have you analyzed any network traffic that included unexpected voice samples? Please, explain yourself.

Those indicators aren't trustworthy. If your phone is configured to respond to "Hey Google", it will do so even when you say it while that indicator isn't lit up.

That doesn't explain how facebook (google's competitor) got ahold of that info, or how people with iPhones are being eavesdropped on.

> That doesn't explain how facebook (google's competitor) got ahold of that info, or how people with iPhones are being eavesdropped on.

It can easily explain how people with iPhones are being eavesdropped on: "those indicators aren't trustworthy". According to this theory, the microphone in an iPhone is active despite the indicator not being displayed. That's identical to the Android theory.

To extend the theory to Facebook, you'd need to assume one of these:

(1) The unreliability of the indicator on Android is a matter of incompetence on the part of Google, and Facebook is exploiting it.

(2) Google has a privileged stealth-microphone-access permission, and Facebook has negotiated for access to it.

Go reverse the operating system and you'll find this is just a paranoid fantasy. Many others have done so already, so you can also just hit up google and find their analyses.

That's not really relevant to the question here. We have this dialog:

A: You know phone microphones aren't secretly taping us because indicators display when they're active.

B: Those indicators aren't reliable.

A: That can't explain how iPhone users would get taped.

Where did that come from?

The indicators are reliable, or do you have proof otherwise?

You seem to have a strong desire to believe that the conversation upthread contains completely different comments than it actually contains.

Everything discussed upthread is irrelevant because the indicators are reliable, or do you have proof otherwise?

Separate microcontroller that can (barely) do one thing well. Any network traffic analysis that indicates they're using this for spying?

If the Tensor security chip works on the Google side the same as Apple there is dedicated hardware specific to detecting the trigger phrases. No sound is processed at a software level until that security chip requests further processing at which point it shows the fake LED. The LED itself is added to the graphics layer in a way that cannot be hidden or obscured.

It is always a question of trust, but after more than a decade of phones they are pretty determined to earn trust. I have not seen any trustworthy reports of a secret always on microphone. Just the occasional recording made after a misheard trigger word or button which is an issue being further secured year by year.

Amazon on the other hand…

Whether it was actually used or just promoted is still unknown as far as I can tell, but the capability was definitely advertised: https://www.sify.com/ai-analytics/active-listening-feature-o...

It's nonsense, only mentioned in some random deck with no technical details, the article says as much. So probably just a hopeful middle manager that let their fantasies get the better of them. Got an actual source?

The accursed Cox Media Group has since released the following statement on their website: https://www.cmg.com/news/cmg-responds-reports-about-disconti...

    CMG Responds to Reports About Discontinued Active Listening Product

    As we stated before, CMG businesses have never listened to any conversations nor had access to anything beyond third-party aggregated, anonymized, and fully encrypted data sets that can be used for ad placement. The information referenced in recent stories is based on outdated materials for a product that CMG Local Solutions no longer sells (although the product never listened to customers, it has long been discontinued to avoid misperception). CMG Local Solutions markets a wide range of advertising tools. Like other advertising companies, some of those tools include third-party vendor products powered by data sets sourced from users by digital publishers and other applications and then packaged and resold to data servicers. Advertising data based on voice and other data is collected by these platforms and devices under the terms and conditions provided by those apps and accepted by their users. This data can then be sold to third-party companies and converted into anonymized information for advertisers. This anonymized data then is resold by numerous advertising companies.

In particular, let me highlight this sentence:

    Advertising data based on voice and other data is collected by these platforms and devices under the terms and conditions provided by those apps and accepted by their users.

To me, this sounds like "We did not spy ourselves, but we bought data from other people who did the spying for us, and that was legal because some user checked a checkbox somewhere".

Watch out, they can also listen to you via the radio waves coming from your dental fillings.

Reply on news.ycombinator.com