So, here's the weird catch-22. Or perhaps, the illusion of choice.
There are two people who are in a photo. One wishes Facebook to use face recognition, the other does not.
Facebook will run face recognition on the photo regardless, but how will they know who wants to be recognized, and who doesn't?
Simply, they have a model of your face already trained. And they recognize you in photos--you have no say whether or not they will. All this option does is hide the notifications when they do recognize you.
Usually these systems are implemented via a generic face detector that first runs on the whole image, to locate all the faces that are present. After the faces are located, they’re matched against a set of candidates (which I would guess is an enumeration if all your Nth order friends [who don’t opt out of this], to reduce computational complexity).
This isn’t the only possible way to do it, but I believe it’s the most common and effective approach.
So it’s not at all unlikely that when running the identifier, they simply don’t match against people who say no to this. They’d never be recognized at all, and the system would just identify their face as an “unknown” identity.
I cannot say for sure whether or not this is how they implemented it, but it seems likely.
> After the faces are located, they’re matched against a set of candidates (which I would guess is an enumeration if all your Nth order friends [who don’t opt out of this], to reduce computational complexity.
Face recognition is usually done by first transforming the face into vector space and than looking for the closest vector from known, e.g. openface[1] (which is based on google 2015 paper "FaceNet: A Unified Embedding for Face Recognition and Clustering"). Facebook itself opensourced[2] a library for efficient search of nearest neighbors in such spaces.
It doesn't seem likely that generating set of Nth order friends (which grow very quickly with N) would be better than just having a global set or a split of global set by some characteristics (e.g. gender, race).
Wouldn't the betterness of using friend networks being reduced false positives? Facebook has 2 billion MAU, which means they have a lot of faces. Filtering down to the 50k or so that are close to you means it's much easier to find the right person, rather than a similar stranger.
> Wouldn't the betterness of using friend networks being reduced false positives?
They are not limiting to friend networks, otherwise they would not catch (out of network) strangers using your pics. One of the use cases touted is "Help protect you from strangers using your photo"
Also finding images you’re in but may have been taken by someone not on your friends list (think events). Eventually probably recommending friends from these images.
Anecdotal, but years ago I uploaded a picture of a friend. He's a bit of a hermit, and doesn't have social media accounts. The picture scrolled by several times that week, and Facebook asked me to tag who this was. (This was perhaps... 2010?) I uploaded perhaps twenty or so photos, but that was the only one they asked about.
I haven't seen this for a while, but I used to get my friends pictures scrolling by. (Not even my own!) "Tag Becky in this photo? Yes/No". Clicking yes confirms what Facebook already knows. Clicking No tells them... what? That they were right, and it is Becky, but I don't want to tag her? What does not responding to the prompt mean?
If you want to fight it, tag them as wrong persons to mess up their training dataset. I am sure they would love that! :) Maybe even start a Facebook group "False positives" where people would do this for fun and to "stick it to the man" lol
I don't use Facebook myself for a few years already, which is paradoxical as they were headhunting me recently to lead one of their ML teams and didn't mind that I quit them long time ago :D But maybe they were just harvesting my own "outlier" training dataset - who knows if I was filmed while onsite and now my gait is analyzed to smallest details including reconstructed skeleton and muscle attachments, my appearance is predicted for the next 30 years under various aging assumptions, maybe later used to fabricate realistically faked video etc. These things are now possible and not that difficult anymore.
Why not? You keep up to date, experiencing world-class hiring process and its requirements (IMO FB's hiring process is the best I've ever experienced, Google's feels one level down), you meet interesting people you might later work together on something cool either there or at another company, you visit interesting places, get some inspirations for your own businesses, interviews might be also a lot of fun (one interview I had at FB was an hour of one-joke-after-another while discussing/inventing architecture of some scalable system; imagine Cryptotinder which is a joke on its own) etc.
It doesn't work that way. They can see that you tag wrong too many people, and you're obviously an outlier, so your data is rejected. The only way you can "win" is by deleting facebook and pushing for better data protection laws.
You won't do it alone, but if most of your friends start doing that, either you mess up their training dataset by introducing a lot of noise, or you all will get flagged as outliers. Both situations are win comparing to learned helplessness one might fall into when realizing the ways how this data can be misused.
Right now the systems in public domain reach 97% accuracy. Facebook doesn't need you, even if you get a million people to try to "trick" the system, you won't do anything. The verification is to fine-tune it.
By the way, you forget that Facebook can simply exclude almost all those who know about this. What are you going to do? Do you think nobody will notice the 10M user fan page themed "Fool facebooks AI"?
It just knows an extra thing about you, that you don't want to be tracked, so it'll hide news about tracking from your feed.
We already know Uber is cherry-picking users it suspects as working for competitors, so precedent is set for what you are suggesting. It's doable obviously, and if it's doable, companies have already either implemented it or having somebody working on it.
You can't undo the past but you can safeguard the future.
Besides, if facebook goes bankrupt with heavy legislation demanding the destruction of user data and not allowing their exchange as assets, they will eventually be deleted, along with their training set and everything else. Your biometrics aren't coming back, but at least only the government will have them.
I'm sure they're on the lookout for that, as well. If they have enough confidence in their model, I'm sure they'd raise an eyebrow at something that only has 30% confidence, say "thanks", and then simply flag my account to not be trusted to train their models.
That still sounds like win. If it gets momentum, they won't be able to recognize any new additions due to them being not trusted. Physique changes significantly roughly every 10 years as well, so if all they have are outdated photos... OK, Deep Reinforcement Learning might be used on reconstructed 3D models from younger photos to simulate aging process, but it might not be very accurate even if looking very realistic.
Similarly to "suggested edits" for pages/businesses. You have an accuracy score as an editor based on how many of your suggested edits get accepted/approved.
I'd be willing to bet my hat that they have a similar scoring for approving suggested tags - don't click them/ignore them complete for a few days and you'll probably start seeing that suggesting a lot less.
Think about it this way: the photo itself already contains all the information necessary to identify you, and algorithms already exist which allow that information to be easily extracted. No privacy setting in the world will change that, short of the "delete" button on the photo itself (or maybe a way to blur your face).
Therefore, this setting logically _cannot_ change what information Facebook has about you; it can only change how that information is displayed.
If you have a book you certainly own all the information in it, but if you are never allowed to open the book and read its content (your brother wrote it but was then terribly embarrased about it and made you swear nobody would be allowed to read it) then how can you put the information to use?
A human might read that book and be unable to control how they react to that information (maybe you might look at your brother differently if you knew what he wrote).
In contrast, a computer could read the book, do a bunch of processing on it, run it through a few neural nets, check your brother's privacy settings, see that the book is marked as private, then erase all its "memories" of the book from existence; the result being absolutely no different from if it had never "read" the book at all.
So even if it were true that "Facebook will run face recognition on the photo regardless", that would be irrelevant as long as it does nothing with the information obtained via that algorithm. And if you don't trust Facebook to honor your privacy setting and do nothing with the information, you shouldn't trust them to not run the facial recognition algorithm in the first place.
Which is why I do not want anybody to make any pictures of me, especially not if they are active on social media. There are exactly two pictures of me online, the one is 30 years old, the other about 10. The quickest way to get me pissed off is to point a camera at me.
I find that a little bit sad. Both of my parents have passed away (fairly young), and they were somewhat camera-shy when they were alive (not for the same reasons you are, but the result is the same). I have barely a handful of photos of them, and that's often a source of disappointment and sadness for me.
I get that rampant data collection is unacceptable, and companies are terribly poor stewards of our personal data, but at the end of the day I have precious little in a visual sense with which to remember my parents.
I think it's entirely reasonable to request that your friends not post pictures of you on social media, though, and get pissed at them when they ignore your wishes.
Good point. It's hard to be sad about something that doens't exist and you can't fathom as a possible technology. Commissioning painted portraits were certainly a thing before photography, but I assume you had to be decently well off to afford that, and of course that's going to capture you in a staged setting, nothing remotely candid.
There have been paintings and drawings of people for centuries. And there are possessions of those people, things that they wore or touched when they were alive.
I'm getting severe anxiety from having a camera pointed at me since this all Facebook stuff started and people posted pictures online. Needless to say, I don't have an account there.
I also take off the moment I see a camera unless the shooter in advance ask me if it's ok and they agree to not put it online. Unfortunately I can't do anything about CCTV but that's not posted online at least.
It's to the point where I avoid social settings with a lot of unknown people and generally stay with my wife and kids. Not that that is a bad think, I'm quite happy to be with them :)
It’s a shame Hyperface has been delayed. I honestly look forward to the day when you see cyberpunks walking around with these on their faces and you can instantly recognize a kindred spirit, if not the identity of same.
My combination of first and last name is unique. Only 23 people have the same last name and none of them have my first.
If I search for myself on facebook there are zero results. But there are probably photos of me on facebook in some form, that I don't doubt. I can't see all cameras..
Yeah, I meant in more in shadow account way. They know who should be there but isn't (at least if you have friends and at least one of your friends gives them access to the phone contacts).
Is this a riff on the 'if you've got nothing to hide you've got nothing to fear' theme?
I just don't think it is ok that large companies are working their way into datasets about every living body on the planet and that is my right.
See also, article 12:
"No one shall be subjected to arbitrary interference with his privacy, family, home or correspondence, nor to attacks upon his honour and reputation. Everyone has the right to the protection of the law against such interference or attacks."
I can't speak for the guy you're replying to, but I'm also curious what harm you're afraid of, mostly so that I can also be afraid of it.
Definitely not the "nothing to hide" stuff as I'm very pro-privacy.
I have a vague fear of some kind of dystopian situation that arises or is influenced by these huge, accurate data sets. I'm curious if you have more specific qualms with it. My vague fear hasn't kept photos of me off of social media, even though I'm okay being a "weird guy who doesn't take photos" if need be.
I know your username and respect your opinion, maybe the grandparent does as well.
I'm not afraid of anything, I just want a modicum of control over who has access to my data and Facebook is one of the companies that I would not trust with anything.
Why does there have to be a scenario attached to this? They've shown time and again that they don't give a rats ass about the people whose data they collect.
The default is not 'on' when it comes to sharing your data, the default is 'off', and even then it should be grounded in need. As long as I don't see the need there is no other reason that I would need to justify my behavior.
On the other hand, there is a very large need for FB to explain its behavior but hardly anybody seems to care. So I'm happy to be the exception.
I don't read that as "no one can take my photo", though. Certainly that language should protect me from people invading my private space to take photos of me, but if I'm out and about in public, I should have no expectation that no photos will be taken of me, and I would feel very wrong asking someone to delete a photo of me in a public place. I would consider that to be infringing on their rights.
I get that photos of people can be a lot more harmful than a casual photographer might assume, but I would never accede to a request from a random stranger to delete a photograph that they just happened to be in. If someone requested that I not take a photo of them, I personally would comply out of a sense of politeness, but I don't believe anyone has a legal or moral obligation to comply when in public.
(But yes, I know, this is just my opinion and preference, just like yours is yours.)
It depends. Tourist taking picture of Dam Square and I'm walking by: no problem. Tourist getting up-close and personal and making a picture where I am the subject of the picture will probably find me very annoying. Friends and family already long ago gave up trying to take my picture for their Facebook pages, the last instance I remember was two years ago in a restaurant in Amsterdam and on first request the image was deleted. People in general are pretty reasonable, whether they have a legal obligation or not is usually trumped by good manners.
As I understand US law, when you're in public you have no expectation of privacy, and you can't do anything about having your photo taken. This is tangental to your point, but does that also mean that Facebook is legally free to ID you in photos which are taken in public? I really don't know.
(FWIW, I'm with you, it makes me crazy when family members post all their holiday photos on Facebook and tag me in them.)
I don't live in the US. And besides that there is a huge difference between just making a photograph and making a photograph with the intent of publication. That's when you get into portrait rights and such, especially if the person you are photographing is the main subject of the photo. Then it is mostly a matter of respect.
For celebrities and politicians there are yet more rules and exceptions.
This is not arbitrary. There is a clear upside: the ability to quickly find all pictures of a person, either by that person themselves or by other people. The benefit provided by this feature has to be weighed against its potential for harm in order to decide whether it is a permissible use of technology or not.
To be honest, facebook should have gathered consent in a GDPR-compliant way: opt-in. But IMHO the feature isn't bad in and of itself. Even recognizing people not on facebook isn't necessarily wrong, as long as facebook isn't building a profile of that person (but merely doing the equivalent of recognizing bananas in all of someone's pictures so they can do a search for banana.)
But given the value provided by the feature, again I ask: what is the harm you're afraid of?
> There is a clear upside: the ability to quickly find all pictures of a person, either by that person themselves or by other people.
That's not a benefit to me.
> But given the value provided by the feature, again I ask: what is the harm you're afraid of?
You're stuck. It's kind of annoying when you ask someone a question, they give you an answer and then you repeat the question again. Either you're going to have to live with my answer or you're going to have to live with not getting the answer that you want.
BTW, I find it hilarious to be lectured by you on HN where your profile is as good as anonymous and mine is stating exactly who I am and where I live.
I believe I understand your point of view (and please correct me if I don't): you don't want these companies to track any data about you because you think of digital privacy as an absolute right.
Now try to understand my point of view: I think treating digital privacy as an absolute right is probably a mistake. First of all it's a denial of how the real world works (where you never have absolute privacy since other people will always know things about you). Secondly, the potential for social data to create value to society is quite large, and absolute privacy means we won't have any of that value. Admittedly, there is a also a big potential for harm, but that's my point: let's be deliberate about these things, not just dismiss them without consideration.
You set up a strawman and then eloquently argue against it.
No, I don't think digital privacy is an absolute right. Obviously some level of online interaction is ok with me otherwise we would not be having this conversation. But I do believe that individuals should have some control over what data of their ends up online and that simply exercising that control should be possible without getting accused of being afraid of specific use cases of that data.
Value creation to society doesn't interest me one bit without an actual example of such value created. The only value FB cares about is the value of their stock and since I'm not a stockholder it doesn't create value for me.
The fact that there is potential for harm would be enough reason for me to be considerate but if even if it weren't I'd still be within my rights to limit which bits of me are online and which are not, it is you that rejects stuff without consideration and it is you that puts the onus on the people exercising their consideration to prove to you beyond some nebulous standard of doubt that those people are not afraid of something when in fact this was never the case to begin with.
I'm a woman, so different situation, but I took most photos of me offline at some point. Later, I began putting up new ones, but it was handled differently. There are more than two photos of me, but there aren't tons. I actually like being photographed, or did in my youth. But I don't like having photos of me all over the internet.
Years ago, I joined an all female Yahoo group that was a spin off from a parenting and homeschooling list I had been active on. I put up a photo of me on my yahoo profile that was in no way intended to be sexy. My long standing friends in this small group of 20 members wanted to know what I looked like and it was for them.
Within 24 hours, I was being solicited by some 19 year old total stranger who lived within an hour of my home who wanted to know if I would sleep with him. My oldest child was 16.
I laughed and my reply clearly signalled that I was laughing in his face and he did not write back. But it did not stop there. Other men would write me and go into graphic detail as to how they would like to spend the next 3 day weekend in bed with me. The only thing they knew about me was what I looked like. That was it.
At some point, I pulled all photos of me off the internet that I had any control over. The only remaining photos were taken by internet friends at a meet-up and were quite unflattering. I later found that a relative who was cyberstalking me while not speaking to me had those photos bookmarked. This was a creepy experience.
There are photos of me out there. Some are not hideous. But I make zero effort to look conventionally attractive. One photo of me online is of me with a shaved head. I am prone to posting photos that are intended to tell people stalking me "Hey, stupid. I am old and ugly and you can stop fantasizing about me."
It hasn't stopped men from hitting on me. But it has been a few years since a total stranger who knew nothing about me other than what I look like has written me to let me know what a pig he is in glorious detail.
I intentionally kept a fairly low profile online for some time. Nonetheless, my super low levels of internet fame still fostered all kinds of weird interest from people who are not healthy and absolutely did not have my best interest at heart. If I had more fame, I would probably be super paranoid controlling. People do all kinds of bizarre things and photos are an excellent way to fuel the weirdness. It makes people feel like they know you, have a personal connection yadda.
I have worked really hard at figuring out how to open my mouth in public and feel reasonably safe doing so. Being penurious with the pics has been an important part of that process. People are quick to objectify people based on a photo. You stop being a real person in their mind and become a symbol of something. It may be a symbol they lust after. It may be a symbol they hate. It may be any number of things. But they will forget you are a real person. They seem to be much less quick to do that based on written comments.
I'm not particularly beautiful these days. I did see myself that way in my youth. My personal opinion of my youthful experiences has changed some.
My mother placed a high value on clothes and I have a good eye for some things and I was surrounded by people who liked me and were good with a camera. An awful lot of so called beauty is manufactured.
I am not comfortable with you characterizing me as a beautiful woman. I have yet to say much about it anywhere on the internet, but will probably blog about it at some point. I have carefully positioned myself to be something other than the stereotypical woman measured primarily by my looks and whether or not men find me hot. Most wealthy, powerful women are happy to use their looks and sexuality as a part of how they make money. I think this is really corrosive to the lives of the average woman.
Beauty is, in some sense, not real. We invent it. And I am not comfortable being accused of that since I make no effort to participate in the societal conspiracy to paint women as good for one thing -- sex -- and to qualify for that primarily based on one metric -- looks. In fact, I go out of my way to sidestep all that nonsense.
Same here. For a while when going out with friends and someone wanted to take a group photo I asked if it was being uploaded to Facebook, and in that case I would get out of the camera sight, but then I realized any photo will end up on FB anyway because most people don't give a damn about their and others privacy, so today I don't allow anyone to take photos of myself in any situation, and ask them not to tag me at any cost should I appear in any photo by mistake.
That way I'll probably just delay a bit the day my photo will appear with full name and personal details: I can't smash all phones of the people taking selfies in a pub or on the beach with no consideration for whoever is sitting/walking/bathing 5 meters from them.
Yes, that is extreme, just as FB complete lack of respect for privacy.
If I had the right connections I'd push for some countermeasures to be added to the phone OSes, such as for example a barcode that disables the camera or covers the wearer's face through image recognition if taken from a short distance: just wear a shirt/badge with that barcode and your privacy is guaranteed: either they move away or point the camera elsewhere.
I uploaded hundreds of stock photos to Facebook and tagged myself (and a few friends) in all of them. I even used multiple photos of the same model to try to foil some attempts to filter inconsistencies. I hope to inject enough bad data into their system that it's totally unreliable for my face.
I have the same reaction/aversion to having my photo taken, for the same reason, but I’ve largely thrown in the towel on this one. The sheer number of cameras out there I have no control of or even knowledge of, plus the constant struggle of having to explain to everyone with a camera phone why I don’t want them to post my picture, plus the growing number of services that upload to the cloud by default... this is an unwinnable battle.
I’ve decided I don’t care if Facebook has my picture. It’s not like they can do anything to me. I don’t use their service, and nothing that goes on on Facebook affects my life.
I'm with you. But I'd also be worried about coming off as the angriest (or most paranoid) guy on earth. I probably already hold the title in my family.
What would be neat is a polite discreet social signal that politely discreetly signaled this to people with a camera. If you did it right, you might also get rich in a Pet Rock sort of way.
I've had exactly one run-in over 10 years or so about this with someone who did not believe me when I told him to delete a picture of me he just took. Other than that everybody seems to understand just fine.
If you "told me" to delete a picture that included you that I had taken in a public place, I would "tell you" no.
If you politely asked, I would generally delete the picture, depending on whether or not there was a reason that I felt I wanted the picture and whether or not there was a way to reproduce the picture without you in it. For example, if Bigfoot had run through the frame while I was taking the picture, I'd be less likely to delete it.
In the country where I live (the US), I have every right to take a picture in a public place with you in it, even if you are the primary subject. I can even publish such a picture on a blog, Facebook, Instagram, etc., pretty much anywhere as long as I am not using it for advertising purposes.
I'd be very interested to learn what "things could get interesting" means - as would the police, depending on the definition of "interesting."
I know only one coworker who understands this concern. No one else seems to care, and the girl I am dating, who is quite technical, has been very suspicious and saddened by me not wanting to put up any pictures of us together. It was basically I relent or something is really wrong, so I gave up. This is the world we live in.
This is a good approach. However this way some positions of power will be left forever locked to you as popularity moves the world (but that exclusion might be a good thing as well if you want to keep sanity/illusion of privacy). Just there are trade-offs.
Why is simply insisting on something that is listed in the universal declaration of human rights something that should warrant a tinfoil hat?
In my experience stuff like this is abused sooner or later, most likely sooner. I see absolutely no reason why I should willingly consent to such abuse. Privacy is important, guard it because if you don't there is no way back.
Looks like you are active on social media (Twitter, anyways) which seems at odds with your extreme privacy stance regarding pictures. Why the distinction? (not trying to attack, just curious)
Twitter has that one pic on it that is a decade old. It doesn't bother me much and it serves a goal. It also matters to me that I have some modicum of control over this and on FB it all gets vacuumed up and thrown together. With this tagging example here as one of the worst excesses.
Which parts of my life go online and which don't is mine to decide.
Your assumptions are not borne out in practice. Others that purposefully take my picture in order to share that on social media does not include total strangers making pictures on the streets.
But if a stranger on the street wants to make a close up picture of me then it could get interesting, so far nobody has been that impolite.
> For the downvoters and flaggers: do you really believe parent asked his question in good faith?
That's not implied by downvoting or flagging you for breaking the guidelines ("Do you always ask dumb questions?"). The thing to do with troll comments is flag, not feed.
That’s an interesting point. Not on Facebook but they know who I am and who I know already (they sent me an invite to join years ago with a list of about 10 people I might know. I knew 9 of them...
Do they have a model of your face already trained, or are they just using a generic facial recognition neural network to compute an embedding vector and then finding identities that are close to that vector in L2 distance or whatever?
How do they discern the difference between someone without an account (ie: a new face), a poor result from the network, and someone who's opted out? Unless the default for everyone is "opt-out", which I highly doubt.
Maybe they use some sort of locality-sensitive hash to efficiently gather a set of (approximate) candidates that are close by in that embedding space, and then use some heuristics to further narrow down who the most probable match is.
If they don't tell anybody who they recognized in the pictures, it should be fine since you will never know? They do a lot more analysis in the background that they never publish anyways
Well your friend took the photo presumably and uploaded it to Facebook. Let's assume that the friend does know about face recognition and does want it enabled for their photos. Surely they have the right?
What about all the face recognition going on in all my Google photos that are automatically uploaded to the cloud? Everyone of the people in my photos has no say in what I do with a photo I took.
I got nearly the same notification, except mine defaulted to off, it said "This setting is off, but you can turn it on any time".
My guess is that they based the setting on another setting. For example there was some photo tagging thing added years ago, and I disabled that. Most of my timeline and tagging settings are set to Custom or require review.
Can't find it in my Facebook settings - can you give us a pointer where to find these settings in Facebooks cesspool of anti-patterns they like to promote as the "settings" page...
There's a "Face Recognition" tab that's new to the settings page since the last time I went through disabling things. For me, it shows up on the left side right under "Language".
For me I had a related setting under the "Timeline and Tagging" tab (set to "No One"), there was no "Face Recognition" tab... until I click "Go to Settings" in the notification, then the "Face Recognition" tab appeared, and the similar setting was removed from "Timeline and Tagging".
It would appear to be the case - I don't see it, but others in Europe do... its maybe also related to local laws (Austria) disallowing such things, and if that is the case I sure am thankful.
No idea. I joined early and didn't really think through some of the privacy ramifications until 2008 or even 2010. Then there are the friends who don't think about them at all, and I assume that Facebook puts some stock in their tags, even if I delete them later.
Same here. Defaulted to off. They could be running an experiment. Roll it out to X percent of people, default half on and see who turns it off, default half off and see who turns it on.
Facebook is notorious for doing A/B testing and feature flipping. It's very possible they are testing defaulting to on vs defaulting to off and measuring how many people turn it off and how many turn it on after the notification.
Echoing this, mine was defaulted to off. I imagine it's correlated to another privacy setting, so for those of us who keep it fairly locked down it will likely default off, although it's entirely possible that it's an A/B test and everyone should be sure to check the setting (Hamburger -> Account Settings -> Face Recognition Settings)
Just adding my name to the rolls of people for whom the setting defaulted to "off." Facebook loves to A/B test these things, which you'd think people would applaud.
I think people are overly hard on Facebook honestly, such that they assume ill intent with everything including assuming it's just default on for everyone.
The fact that they notified you means it wasn't exactly silent. I saw this notification and turned it off. I wish it hadn't been opt-out and had been an opt-in thing instead.
Was it a push notification? Or did you have to open the app to see it?
It's silent unless there was a proactive message to you from Facebook that your face will start to be recognized. This facial recognition happens even when you're not using the app.
So Facebook is waiting for you to take action before telling you anything. That is not proactive, it is silent.
What if you don't go on Facebook any more but have an account? You'd never know this was happening.
It may be silent for some people and you're intentionally ignoring that. For someone who has an account but does not regularly log in, this change is silent. Yes, you saw the message, but that was by chance. Many others will have this feature silently enabled without their knowledge.
"I didnt get any mail. I also havent checked my mailbox for packages."
"I didnt get your email. I have not logged into my account lately."
It seems like semantics. You expect facebook to send an EXTERNAL notification to an operating system like ios/android or to an email provider like gmail/hotmail. Other people consider it not silent for facebook to make sure it shows up in their newsfeed.
Good for you but I don't open the app too, I wouldn't know what Facebooks does with the app but I guess Mark Zuckerberg enjoys The Hitchhiker's Guide to the Galaxy.
How does the opt-out work? Is it simply turning off the notifications or is it also not storing the match when it recognizes your face? Clearly it doesn't know who you are (as an opt-out participant) until it scans the image.
I think this is notification is sort of the opposite of what people think it is.
Facebook has had facial recognition and tag suggestions based on facial recognition for a long time. For example this [1] article from June 2017 explains how to turn it off.
This new notification is about a global on/off switch that disable all facial recognition, because Facebook is going to start doing more than tag suggestions with facial recognition [2][3]. For example they are trying to detect when people are impersonating you online.
The global face recognition setting defaulted to off for me, probably because I had tag suggestions disabled already.
More than a decade ago Mark Zuckerberg said that anyone who gave him a photo of themselves was a "dumb fuck". His entire plan this whole time has been to do exactly this kind of thing (edit: because he thinks it is profit-maximizing), and he thinks you're a dumb fuck (just like me) for uploading your photo to let him use in the maximum-scary-surveillance way.
We should all just get off Facebook. The problem is that you can't leave even after you "delete" your account, and they have your photos tagged and matched to your email even if you never signed up anyway since all your "dumb fuck" friends gave Mark Zuckerberg their email passwords too - so even if you never sign up, your friends already ratted you out and gave up your email and photos anyway.
This is why I think the 'right to be forgotten' legislation might actually be a GOOD thing. Sure, it's some extra work when designing a new system, and lots of legacy systems require expensive rework, and a few twats will abuse the right to have Google redact journalistic information about their personal misadventures from the public sphere. But it's still my private data, so I should have the right to forbid companies I don't trust from storing and processing it. And they should be legally forced to oblige and have processes in place so that such request can be truly honored.
Yeah, if you think Facebook would EVER honour a system like that then I have many, many bridges in stock for you :)
To most people, flicking a bit from 0 to 1 is enough to convince them your data is "deleted", however, in the database, what people never see, is that it the column name is actually "hidden".
It's not that we naively trust him, we just believe the creepy data gathering is the price to pay for that warm feeling of being connected with pseudo-friends and kinda-acquaintances.
It's not just creepy data gathering. It's specific, targeted emotional manipulation. Facebook runs "experiments" on its users to intentionally make them feel sad or depressed. They publish these works, proud of their accomplishments, and tell us that we agreed to be emotionally manipulated into a dark depression because we agreed to the Terms of Service when we signed up. Facebook believes you gave them your informed consent to fuck with your emotions because they believe you read their TOS and they believe it is morally okay to do stuff like this.
No, it's not worth those warm fuzzy feelings you get, because half the time it won't be warm and fuzzy but actually, it will be a depression that Facebook intentionally put you into, just to see if it could.
Facebook is not worth it. Email lets you stay in touch just fine.
And people will have the FBI going through their mail because an algorithm written by someone who lived by "move fast and break things" flagged them as a commie (or whatever the equivalent is 30yr from now) over some combination of things they posted in high school.
If you spread incompetence around just right the end result is approximately the same as malice.
Here’s the press release announcing the new control from December.
> People gave us feedback that they would find it easier to manage face recognition through a simple setting, so we’re pairing these tools with a single “on/off” control. If your tag suggestions setting is currently set to “none,” then your default face recognition setting will be set to “off” and will remain that way until you decide to change it.
True, there is nothing stopping them from running this in the background to make additional linkages between people and places to sell to third parties without having to share where those linkages come from.
I'm imaging a Probability Zero or Black Mirror short where Facebook insists some scammer is using your face and attempts to shut them down. They are unable to log into services because they used Facebook login for convenience. In the end, it turns out the imposter randomly had an identical face and was perhaps a long list twin.
I have to assume that this UX is intentionally awful:
Go to Settings.
Go to More.
Tap Privacy Shortcuts.
Tap More Settings.
Tap Face Recognition.
Tap Do you want Facebook to be able to recognize you in photos and videos?
Select No if you don't want to let Facebook recognize you in photos and videos.
Also, this set of steps doesn't work on the desktop browser. I don't see More or Privacy Shortcuts in My Settings, only in the ? menu at the top (clicked by accident). Once you go to More, I don't see any section for Face Recognition anywhere on the page.
I don't see anything but a divider between "Language" and "Notifications", and for me the only "face" anywhere on the "Timeline and Tagging" page is "Facebook".
Even on mobile the set of steps they published doesn't work. "More Settings" leads to a generic settings page with no "face" as well.
From other comments in the thread, it sounds like there are at least 2 different locations that they put the option. It's either region/country-specific, or an example of A/B testing, I suppose.
I wish I could activate it once to see all the previous photos of me that aren't tagged, then turn it off so it doesn't do it in the background. Not that its any safer, but it would make me feel better about it.
As the top comment suggests, if Facebook runs face recognition on pictures of you and those pictures contain other people, there is no guarantee that they have opted into the program as well. You wouldn't just be making a choice for yourself.
Well, either Facebook doesn't have training data for the opt-out users, or they do, and they use it to ensure that those users aren't publicly tagged (my paranoia says that they're still tagged internally, regardless of the option that's selected).
Either way, it would only need to identify those users who have actually opted in.
Did anyone else notice that FB mentioned accessibility as one of the reasons that people should opt-in to face recognition? I wonder whether this was something that the FB Accessibility team actually asked for, or whether this is just a do-good sounding excuse for a program that FB wants to (clandestinely, apparently) launch.
The feature is "opt-out" in the sense that they based the default on some other privacy setting. It was disabled by default, for me. But I've been fairly diligent over the years, going through and disabling the things that I didn't want.
I'm OK with this, under the condition that Facebook only uses it to alert you of photos you likely appear in, i.e. those two first examples of what it can do.
I would not allow them to show those results to other people.
Google Photos' was only within your account, and didn't assign any of the faces to you unless you manually assigned it as "me". Later they included the option (via opt-in) to have Google suggest that your friends share with you photos that have you.
Facebook's is platform-wide, and you don't have the option of tagging faces but not assigning them to you or to specific people.
As an app developer, I don't want to say this is impossible, loading and abusing private libraries has been done in the past. It just seems so unlikely. And it seems like if proved (decrypting network traffic from mobile device?) would destroy Facebook's image w/ customers.
I think more than likely the talking about a certain subject makes the ads stand out more than they would have. It's like, researching to buy a new car, all of a sudden you see that car everywhere on the road. It was there before but you just didn't notice it.
That's my take on it. Again, it's not be impossible. I just think it would be highly likely for facebook of all companies to directly snoop. (Maybe by proxy of other apps?) </TinFoilHat>
> And it seems like if proved (decrypting network traffic from mobile device?) would destroy Facebook's image w/ customers.
Would it, though? Facebook gets caught with its pants down standing in front of a row of bent-over users pretty regularly, and for the most part those users seem happy or naive enough to stay where they are, or at best just migrate to other Facebook-owned services like Instagram/WhatsApp.
What are some good examples of that? Specifically something that has malicious intent or an abuse of some TOS. I can think of a number of examples from companies like Uber. But maybe I missed something from Facebook.
I've notice odd things like this too. I was visiting my grandma and a commercial came on for the Hover Cover, some silly thing with magnets. We spent maybe 5 minutes making fun of this product out loud, and when I got home that evening Facebook had a Hover Cover ad to show me.
They followed up with that person from that episode, and they all determined that it most likely wasn't microphone listening, just cross platform tracking with instagram.
Just tonight we briefly mentioned garage doors and remote garage door openers, and we got an ad on FB for remote car openers. We haven't looked up any of these things online or anything else, it was a brief discussion that lasted all of 1 or 2 minutes, tops. It's definitely not cross platform tracking, it's got to be microphone listening. I advised her to uninstall FB from her phone, because this is getting obnoxiously creepy.
The episode was really good and stated that Facebook has at least 55,000 data points on any given individual, and when you have that kind of metadata on someone, it seems like parsing voice data would be inefficient and expensive in comparison to inferring information from those data points.
This is why people should forget about attempting to block ads and start attempting to actively ruin the analytics.
I've started to click random links and garbage ads just to mess with the bulk collection of data. It probably won't make a dent, but hopefully it scews a few of those data points
This is a pretty bold claim to make without something more substantive than this anecdote. There's too much opportunity for coincidental consumer profiling to be the root source of what you suspect is something more malicious.
I've seen very odd events as well that have convinced me this is happening.
Went to a friends house and ending up discussing the most random, very specific topics. 2 of the 3 topics popped up in ads on the way home an hour later. Never searched anything related on my phone prior to then.
It happened to me at work recently. I thought this was fake too but my zipper broke on my backpack at work while my coworker and I were walking down a long hallway between offices. We spent the whole walk talking about backpack zippers. When we got to the other office and I looked something up on facebook I had ads for backpack zippers. It was so creepy and neither of us had searched for anything between the conversation and when it showed me the ad. Listening to my mic while I wasn't even in the facebook app was literally the only way they could have known to show me those ads.
There are two people who are in a photo. One wishes Facebook to use face recognition, the other does not.
Facebook will run face recognition on the photo regardless, but how will they know who wants to be recognized, and who doesn't?
Simply, they have a model of your face already trained. And they recognize you in photos--you have no say whether or not they will. All this option does is hide the notifications when they do recognize you.