back to article Image-rec startup for cops, Feds can probably identify you from 3 billion pics it's scraped from Facebook, YouTube etc

Here's a roundup of news beyond what we've covered already in the world of machine-learning. Huge facial-recognition database scrutinized: An AI startup called Clearview claims to have a database of more than three billion photos, scraped from Facebook, YouTube, Venmo, and other websites and services, and all used to train a …

  1. Lusty

    Legality

    "Clearview says it's using publicly shared photos, and thus isn't doing anything wrong."

    They are incorrect. This is the very definition of personally identifiable information - they are actually using the data to identify individuals. That suggests GDPR should give them a very heavy kicking over every instance of EU citizen data held.

    It won't, because this is too useful to law enforcement. Best case is this gets pulled inside an agency instead of being a private company. Then GDPR won't apply anyway.

    1. Pascal Monett Silver badge

      Re: Legality

      Clearview is quite incorrect indeed. There's this thing called decency, where it is not because you can do something that you should consider yourself justified in doing it.

      You can follow someone in the street. If you do that for too long though, that person is going to get cross at you. Arguing that you have the right to be in the street does not change the fact that you do not follow people. Here, it is the same thing. You can look at a picture. That does not give you the right to copy the picture and monetize it.

      Unfortunately, the law is not going to react on this point.

      1. Khaptain Silver badge

        Re: Legality

        Pascal, do "decent" people really need to publish their lifes details and videos online in public forums ?

        How can people expect privacy when they are baring themselves publically to one and all ? It is a very ironic situation ?

        No, it does't mean I agree with Clearview, it's just that the data was laid out before them and they were clever enough to scoop it up and monetire from it, and I really don't imagine that anyone is going to change this possibility any time soon. It certainly wont be Google or Facebook that change the right to publish publically, they are after all making "massive" amounts of money from it already..

        1. Anonymous Coward
          Anonymous Coward

          Re: Legality

          Yep. Protection and defense in depth.

          These "but I need to use facebook" people often forget, a lot use it in private/group only mode. Those I feel sorry for when scammed/data mined. Those never asked for this. However, those being too risky and irresponsible, are not necessarily at fault, but are walking on thin ice.

        2. Luke McCarthy

          Re: Legality

          I think the problem is most people don't think of it as being public - they are sharing photos with their friends and family, and don't expect them to be automatically scraped and added to a database for ulterior purposes. They fact that this is technically possible, most people are unware of. Social media is a very new thing and people have yet to really adjust to the technology or understand its deep implications. Suddenly millions (billions?) of people are as public as politicians and celebrities, but they don't really have the tools, power or ability to manage this level of mass exposure.

          1. Khaptain Silver badge

            Re: Legality

            "Social media is a very new thing and people have yet to really adjust to the technology "

            What ? It's been around for almost 10 years and with all the scandals that we are all aware of using ignorance is simply not an acceptable excuse.

            "or understand its deep implications."

            Why do they not understand, simple, it's because they refuse to accept that Free is not Free after all. We have be told countless time that we are the product, this is nothing new..

            "But they don't really have the tools, power or ability to manage this level of mass exposure."

            Yes, they do, they either choose to keep the data private and not publish to "public" sites like Facebook, Twitter, Youtube etc.

            If you wash your laundry in public then it should only be expected that the villae knows what colour of underwear you own.. No one needs to tell you this, it bloody obvious...

            Publish silly content to a massively public website like Youtube and suddenly you they don't realise what public means.. Please explain to me why suddently people become unaware of what they are doing...

            1. Charles 9 Silver badge

              Re: Legality

              "If you wash your laundry in public then it should only be expected that the villae knows what colour of underwear you own.. No one needs to tell you this, it bloody obvious..."

              I put it another way. If you don't want anyone to know your birthday (or even come close), don't throw birthday parties (I don't--I hate pomp and circumstance).

        3. Anonymous Coward
          Anonymous Coward

          Re: do "decent" people really need to publish their lifes details

          while I have the same revulsion towards FB and social media in general, as I do for other bottom-feeders (and there are plenty in this world, all those, from the smallest to the giants), I think calling into question "decency" of those people who FEEL they NEED to publish their lives' details goes way too far. Naive, uninformed, careless, vulnerable, somewhat synonymous with clueless idiots, yes. Even harsher, narcissistic, hypocritical (and then some), this applies to many, if not most FB users. But I don't think FB usage goes along with "decency" or lack of (whatever "decency" may actually mean), or that this trait is a common one for FB users.

          p.s. naturally, another key word is "need", and there's a 0.000001% of those who "need", but then, definition of "need" is likewise, very broad, no less than the definition of "FREE!!!!".

        4. moiety Silver badge

          Re: Legality

          "How can people expect privacy when they are baring themselves publically to one and all ?"

          How about normal people who some cunt has tagged on Facebook or whatever? It's not just their own privacy that these dipshits are violating.

          We taking bets that every one of the 3 billion on the database put themselves there?

          1. Graham Dawson Silver badge

            Re: Legality

            Exactly this.

            My face and name are all over Facebook, though I don't have an account there. Other people tag me in photographs they upload and share, third-party pics shared with them eventually start suggesting me as a recognised face and that's that. I'm filed, stamped, briefed, debriefed and numbered.

            Short of demanding people never photograph me, never leaving the house, and wearing a mask in public, there's nothing I can do to stop that process, not millions of others that also have no account.

            1. Khaptain Silver badge

              Re: Legality

              If all my friends spent their time on facebook uploading and tagging photos then I would consider changing my entourage...

              1. moiety Silver badge

                Re: Legality

                Takes one person one time.

            2. spold Silver badge

              Re: Legality

              Hmmm you should be able to send Facebook a Data Subject Access Request under GDPR, using a photograph as your identifier (photographs are personal information and uniquely identifying), and ask them to delete all data about you including other images.

              I'm quite tempted to try this...

              1. Graham Dawson Silver badge

                Re: Legality

                Now you mention it, so am I, but at the same time, I'm pretty sure facebook would just save the picture, store my e-mail, tell me they deleted everything and not actually follow through.

        5. DavCrav Silver badge

          Re: Legality

          "they were clever enough to scoop it up and monetire from it,"

          Given GDPR, you mean to say they were criminally minded enough etc.

        6. Snake Silver badge

          Re: Ignorance

          "Pascal, do "decent" people really need to publish their lifes details and videos online in public forums ?

          How can people expect privacy when they are baring themselves publically to one and all ? It is a very ironic situation ?"

          The problem is that the public has been brainwashed into believing, the warning, that Orwell-style 1984 will come about exclusively be government intrusion. As this article proves, the only reason that people draw attention to it is because the government's involvement, only now, is coming to light - note that, quite fundamentally, nobody had any issue with this large enough to prevent it as it got enough funding and technical backing to become a hard reality.

          So as long as it is perceived as being of private interest, nobody put a halt to it (Facebook well included). But now, now that the government is taking advantage of it? Holy cow, we need to STOP IT!!

          So people have been told not to worry about things that are handled by Big Business, they'll look out for you (seen the new Apple YouTube ads about privacy? Apple is looking out for you! ...But only until it's no longer convenient).

          So people have brought this against themselves. I have *no* pictures of me that are not anonymous - even my *ah-hem* adult acting used a synonym.

      2. John Brown (no body) Silver badge

        Re: Legality

        "You can look at a picture. That does not give you the right to copy the picture and monetize it."

        I wonder how many of those photos scraped up by Clearview are publicly published but owned by Hollywood stars and studios? I suspect it may time to buy some popcorn. As you say, it's one thing to look at a publicly published photo, it's quite another to make a copy and yet another thing then use that copy for commercial gain.

        IIRC under UK anti-porn laws, having a digital copy is classed as "making".

        1. Khaptain Silver badge

          Re: Legality

          "As you say, it's one thing to look at a publicly published photo, it's quite another to make a copy and yet another thing then use that copy for commercial gain."

          And yet Google are doing exactly that..

    2. ChipsforBreakfast
      FAIL

      Re: Legality

      I'm willing to bet this particular company is very much located on the far side of the pond from Europe and considers itself sufficiently insulated from such minor inconveniences as GDPR.

      In theory the reach of GDPR is global, the real question is do any of the various regulatory bodies involved have the balls to even try to make that case.

      I won't be holding my breath.

  2. Khaptain Silver badge

    Self Worth and Esteem

    Who is truly to blame here :

    Clearview for taking what is essentialy public information, which it is by the fact that the uploaders choose to put the info and all the details on public forums/website. All of us have the same possibility, in that we can scrape the same info and use it as we see fit. Clearview are just collecting what is available to all and sundry... I too can browse the videos and extract a lot of info.. Should I also be held in contempt of GDPR laws, of course not...

    or

    The uploaders for publically publishing their own and also "other" peoples lifes and all their details..... Personally I do not accept, nor want, my friends or anyone else to publish my activites on the web.... This is what GDPR truly covers, the right of the individual to decide for themselves...

    Those that publish the videos have already agreed to Google terms, the videos on Google, Facebooks sites become Google, Facebooks property to see as they do fit.. This is the unfortunate consequence of "free"....

    Love it or hate it, if you don't want to become public then don't go public... Youtube and Facebook are a bees nest full of nasty.....

    And I can't believe for a moment that Google and Facebook don't already have the same, or even more data.

    1. Stork Silver badge

      Re: Self Worth and Esteem

      So, you (or your kid) is in a sports team and a photo is shared in a group, and then by a member shared outside the group. Bescouse of lack of thought (remember, about half of people are thicker than average).

      From my memory, DB et.al. does not exactly try to stop that

      1. Khaptain Silver badge

        Re: Self Worth and Esteem

        "So, you (or your kid) is in a sports team and a photo is shared in a group, and then by a member shared outside the group"

        This would be the same scenario as the local newspaper publishing a similar photo and with the names of all the players in the subtitle. There would be enough "public" information as to be capable of identifying who is who. Dates, times , locations,all team members and any special achievement that day. Once it's public, it's public.

        So in the future it is important to make a request that no images of you, or your kin, go public without complete agreement from all parties concerned...

        if someone chooses to publish without your agreement then they are breach of GRDP laws... Otherwise it's your reponsability.

    2. Chris Fox

      Publication does not entail a grant of all rights

      "I too can browse the videos and extract a lot of info.. Should I also be held in contempt of GDPR laws, of course not..."

      Actually, depending on exactly what you are doing with videos, and images etc., and how you are processing them, you could indeed be in breach of GDPR, and possibly in breach of the owners' copyright as well. By itself, the act of publication does *not* mean that third parties have been granted all rights to the data, and does not remove any obligation to seek permission on matters governed by the GDPR, or by copyright.

    3. DavCrav Silver badge

      Re: Self Worth and Esteem

      Who is truly to blame here:

      Me, for stealing your organs, or you, for having such a juicy liver?

      "I too can browse the videos and extract a lot of info.. Should I also be held in contempt of GDPR laws, of course not..."

      If you collate it into a database of PII of random people (not, for example, your family) then yes, you have committed a crime. And it should be illegal to keep a database of people's information that they did not give you.

    4. John Brown (no body) Silver badge

      Re: Self Worth and Esteem

      "Love it or hate it, if you don't want to become public then don't go public... Youtube and Facebook are a bees nest full of nasty....."

      Someone owns the copyright to those images. It's not really all that different to taking a publicly broadcast TV show or film and making your own copy. It's a grey area if it's for personal use, but once you start either disseminating it or using it for commercial gain, then you on the wring side of the law in most jurisdictions, including the USA. The problem is that it's probably a civil matter and down to the copyright holder to try and enforce their rights. On the other hand, maybe the scale makes it criminal matter? IANAL so maybe someone can jump in and clarify.

  3. werdsmith Silver badge

    My brief dalliance with social media ended many years ago, but the only photo I used was of a horse. So I'm not quite sure what FBI will find on me.

    Actually, I am sure, I've been through the ESTA process anyway.

    1. tiggity Silver badge

      @werdsmith Indeed

      Only chance of an image with me being in it anywhere on social media is if someone else has posted one

      1. Anonymous Coward
        Anonymous Coward

        Gah.

        You just reminded me. The only online photos of me are from work. Duh!

    2. Warm Braw Silver badge

      My brief dalliance with social media ended many years ago, but the only photo I used was of a horse

      Well, if you're going to post a dick-pic, post one that gets attention...

      Which leads to the question of how web-scraped photos are going to help identify people with their clothes on.

    3. Fruit and Nutcase Silver badge
      Coat

      but the only photo I used was of a horse

      "Mister Ed"?

  4. Andy The Hat Silver badge

    Whether the weather ...

    "Researchers at Google have trained a neural network to predict whether it’s going to rain up to six hours ahead of time by analyzing radar images."

    Millions of pounds of hardware to look at the rainfall radar and tell me whether it will possibly rain in a little while ... when I occasionally do this it takes a few clicks of a mouse, mark one eyeball and a few seconds of slideshow met office images ...

    However, in general, Lucy Verasamy does it for me ...

    1. werdsmith Silver badge

      Re: Whether the weather ...

      I don't think it needs millions of $$$ of hardware these days. A few decent Nvidia chips is plenty.

      Yes, Lucy Verasamy is a really really good weather forecaster. Exceptional in fact.

    2. Dabooka

      Re: Whether the weather ...

      I've been wondering of late about the effectiveness of modern weather prediction over, say, what we used to do back in the 20s and 30s (1920s that is...).

      Last Tuesday I should have been drowned from 2pm onward as a thick band of severe rain moved north. Nope not a thing, or at worst light drizzle for a small amount of time. Yesterday was meant to be 0 or below. Nope I'm outside working on the car.

      Okay hardly scientific, but the point is that I wonder that despite the great amount of cash and computing we're still actually a few % points better at predicting it.

      1. DavCrav Silver badge

        Re: Whether the weather ...

        "despite the great amount of cash and computing we're still actually a few % points better at predicting it."

        Welcome to the world of exponential growth.

  5. The Man Who Fell To Earth
    WTF?

    Who knew?

    Today's Pets.com is to take a few weeks learning Tensorflow (or equivalent), start an "AI" company, and get snapped up for tens's of millions per FTE.

  6. Version 1.0 Silver badge

    Cat memes will save the day!

    So the Feds will now be able to identify everyone who is not a cat lover.

    1. Craig 2
      Joke

      Re: Cat memes will save the day!

      Forget the Feds! Eventually, cats will be able to identify everyone who isn't a cat lover, then we're REALLY in trouble :)

  7. Oh Homer

    Sabotage

    I can think of a good way to spoil the AI harvesters fun, ironically using AI technology against itself.

    Use a Deep Fake generator to create an endless iteration of modified selfies, merging them with photos of other people's faces to produce hybrids, then upload them as profile pictures and identity-tagged posts to every social network that you suspect has ever had your photo. And not just once, but keep renewing your fake profile photo, and keep spamming these sites with fake selfies, basically forever. Oh, and use TOR or a VPN so this activity can't be linked to the same IP.

    If enough people do that often enough, eventually these AI databases will become overwhelmed with so much fake data that they will be rendered essentially useless.

    Just a thought.

    1. Charles 9 Silver badge

      Re: Sabotage

      Or we simply run out of data allowance. It's not like every ISP customer in the world has the pipes to outpace the likes of Google, who owns their own global fiber network. Bet you they're already working on a DeepFake detector already for their own sanity...

    2. MalIlluminated

      Re: Sabotage

      Seems like a very good idea. Regardless whether other people do this, I gain a personal reward for ruining their database about me. It might be selfish of me, but provided the system can't identify me or anyone I care about, I'm bothered less by it's existence. I have a limited amount of indignation to go around, and there's plenty other problems clamoring for their share.

  8. Robert Carnegie Silver badge

    Now,

    Whenever Doctor Who goes anywhere in space and time after the year 2020, they will know straight away that it's him/her.

    ...until 5045 or whenever that s/he creates and distributes a virus that encrypts him/her, or something.

    …like, an actual virus.

    (Possibly an intelligent and friendly one.)

  9. TheRealRoland
    Devil

    Remember Peter "Kill all Gawker b/c mah privacy" Thiel? Meet the financial backer behind Clearview.

    1. DavCrav Silver badge

      Wasn't he the vampire? Or is that some other Silicon Valley ghoul?

    2. Anonymous Coward
      Anonymous Coward

      "Meet the financial backer behind Clearview."

      And not just Clearview.

      As well as the usual stuff familiar to most of the public, there's also Palantir.

      It's not a good picture, except maybe for the regular Bilderberg attendees (including yer man Thiel).

  10. SVV Silver badge

    Clearview contacted its police users to ask if they were talking to the media

    So the company is actually monitoring police activities then......how is that allowed, or possibly even legal? Does the usage licence contain stuff about not being able to use the system for work with people the company doesn't like, such as journalists investigating the ethics and legality of the whole setup?

    It's all way too Black Mirror for my liking, imagine trying to take a train journey or go into a shop at some time in the near future, and the door closes and refuses to open for you because you have run afoul of the greatest crime the control system knows : "Not Recognised".

    I only hope that there will be a final twist in the tale for the privacy obliterating gazillionaires, whereby they themselves get eaten by their own monstrous creations.

  11. Il'Geller

    Artificial Intelligence is a database containing structured texts. Face recognition is an algorithm.

    1. John Brown (no body) Silver badge

      ...and yet intelligent humans can recognise faces almost from birth. It appears AI has a hell of a long way to go to assimilate those facial recognition (currently poor) algorithms. I suppose your AI could have a go at identifying faces created in ASCII art since that's a textual format.

  12. Uplink

    Insane Clown Posse

    How well does it work with Juggalo makeup? Expect a steep rise in users if this takes off.

    1. Charles 9 Silver badge

      Re: Insane Clown Posse

      Even before that, what about KISS fans? They wore lots of face makeup, too, to mimic the band.

  13. RLWatkins

    More AI Bafflegab

    Is "Xnor" pronounced "snore"?

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Biting the hand that feeds IT © 1998–2020