Creepy T-shirts designed to baffle Facebook facial-recognition software

Filed Under: Facebook, Featured, Privacy

Facebook logoFacebook loves your face.

It loves your face so much, it bought the Israeli start-up for its facial-recognition technology in 2012.

It loves your face so so much, it uses this facial recognition - based off photos you're already tagged in and soon your profile picture - to make it easier for your friends to tag you in photos.

Yes, Facebook definitely has, as Wired's Kyle VanHemert put it, a mission "to secure its status as the world authority in who knows who - a constant, lumbering quest to improve its advertiser-serving 'social graph'."

How do you fight back? VanHemert suggested raising a stink with Facebook over the proposed policy changes, changing your profile picture to that of your dog, or - this option just in - buying a T-shirt printed with creepily distorted faces of celebrity impersonators, designed to give Facebook's facial recognition technology a migraine.

Obama tshirt. Image courtesy of Real Face.The garments - dubbed the "REALFACE Glamoflage" T-shirts - were designed by Simone C. Niquille as part of her* master’s thesis in graphic design at the Sandberg Institute in Amsterdam.

The shirts are custom-printed and sell for around $65.

The prints feature distorted faces of celebrity impersonators - Barack Obama, Michael Jackson and others - with the aim of creating an easy way to befuddle Facebook's pattern recognition algorithms, Niquille told Wired:

I was interested in the T-shirt as a mundane commodity... I was interested in creating a tool for privacy protection that wouldn’t require much time to think in the morning, an accessory that would seamlessly fit in your existing everyday. No adaption period needed.

The project was inspired in part by the "ugly T-shirt" envisioned by William Gibson, VanHemert writes.

Gibson's imagined T-shirt hides the wearer from CCTV surveillance, which in turn is similar to a type of camouflage used by ships in World War I - dazzle camouflage - in which the ships were covered in conflicting geometric patterns meant to scramble their speed, range, size and heading.

The shirts' strategy is the same, Niquille told Wired: they won't hide you, but they will mess up whoever's watching you:

They won’t keep your face from being recognized, but they will offer distraction.

Michael Jackson. Image courtesy of Real Face.How well they work also depends on how tightly they're worn, she says: the tighter the fit, the better facial recognition will be able to recognize the faces in the fabric when analyzing photos of the wearer.

Niquille's thesis, titled FaceValue, is about what the designer calls "the (human) face's value in a time of rapidly mutating standards and techno norms" - presumably, that includes biometrics, privacy and pattern recognition.

I say "presumably" because her writing tends to the prosaic rather than strictly expository.

An example:

FaceValue is what you and I will have left after CCTV gets hooked to the Facebook databank and parents look younger than their children. FaceValue is what will make a Britney lookalike earn more than Spears herself. FaceValue is what will remain after contemplating the appropriate nose for the day. FaceValue is what you’ll consider while putting in a new chimplant order at RapidAesthet3d the morning after. FaceValue is what stares back at you and me, relentlessly reflected in the surrounding screens as they fade to black.

In fact, the shirts are only one part of her thesis, which also encompasses two other projects: FaceBay, an online marketplace to buy and sell visages, and FaceValue: Accessories, which examines ways in which technology could enhance or augment a user’s physical face - from contact lenses that emitted an image-obfuscating reflection triggered by a camera’s flash to the Chin+, a 3D-printed prosthetic that offered a more prominent jaw, optimized for FaceTime or Skype video calls, as Wired describes it.

Is this a good way to fend off facial recognition?

Will you be ordering a T-shirt? Let us know in the comments section below.

Oh, and if you're on Facebook and want to hang out with people who use their faces to eyeball Facebook's every last move, join Naked Security's Facebook page.

*Wired's article uses both male and female genders to refer to the artist, and as of the time this article posted, I hadn't been able to determine which is correct, so I went with "she".

T-shirt images courtesy of Real Face.

, , ,

You might like

14 Responses to Creepy T-shirts designed to baffle Facebook facial-recognition software

  1. Deirdre Patterson · 733 days ago

    FYI - The title of your article has a typo:

    "Creepy T-shirts designed to baffle Facebook facial-reconition software" should read: "facial recognition".

    • Anna Brading · 733 days ago

      Thanks Deirdre. We've now corrected it :-)

      • Guest · 732 days ago

        As an FYI, the URL still has the typo.

        • Paul Ducklin · 732 days ago

          If we were to edit a URL after making it live, then any links to the old one would break. So we'd need to keep the old one anyway and redirect it to the new one.

          As a result, once we've published URLs, they are pretty much locked in and we rarely change them.

          (Now you'll be playing "spot the URL typo" on this and other sites...I know I do :-)

  2. John Juliano · 733 days ago

    This seems a little weird and unnecessary ("They won’t keep your face from being recognized, but they will offer distraction."). Isn't it much simpler than that? I just tag random people and inanimate objects (blocks of cheese, cartoon characters, etc.) with my name and disallow (or remove) tags with my name on others' pictures (manageable, since I'm about 20 years too old to be a party animal, so this happens maybe once every couple of months). There are no correctly tagged pictures of me on Facebook, and most of my profile information is designed to decieve (no, there is not a "Yastrzemski Hog Care and Meat Processing Vocational High School", despite my claim that I attended it).

    That's all it should take. Right? Right?? (You guys would know better than I would...)

    LinkedIn, however, presents another problem entirely, since the info there has to be correct. Haven't figured out kluges around that yet.

    • markstockley · 733 days ago

      My guess is that you'll be ahead of the game until you aren't and you won't know when that moment is. Who knows what rules are at work today, never mind tomorrow.

      I wouldn't assume that facial recognition software can't recognise that a table isn't a face and disregard it for instance.

      Also I would (and I am conjecturing wildly) assume that Facebook is drawing on many more sources of information than how you label things. It is not difficult to imagine a fairly simple algorithm along the lines of "the face that appears 2nd most often in photos belonging to person X and is male is most likely her significant other who we know is called Jeff because that name appears in proximity to hers most often in posts belonging to friends she met since she was married".

      And that conjecture is based on a fairly simple set of relationships I can think of. 'Big Data' is all about using a large database to discover meaningful non-obvious relationships.

      The amount of data you control on Facebook is miniscule vs the amount of data they have. They have massively detailed information on a billion people that they can data mine and use to develop rules about the relatedness of things. They can then apply those rules to you and your loved ones and I would be surprised if you were able to second guess those and defend yourself against them.

      And remember they're capabilities are a moving target - you have to get better at hiding as they get better at figuring things out.

      • John Juliano · 732 days ago

        Thanks - that's about as comforting as I expected. But then again, a t-shirt with random faces on it will do about as much good as my strategies, eh?

  3. Islander · 733 days ago

    WEll, if you're concerned about that sort of thing, here's a whopper: "...they will mess up whomever's watching you..." Should be "whoever," of course, since it's the subject of the clause, "whoever is watching you."

    • Lisa Vaas · 733 days ago

      I couldn't make up my mind about that, given that the who/whom is the object of getting messed up.... as well as being the subject who's watching you. You can mess *me* up while *I'm* watching you, right?

      So maybe avoid it entirely by saying "xyz will mess up whomever you like, if a given who is watching you."

      Because saying it in that Dr. Seuss-like way wouldn't be awkward at all! :-)

      At any rate, I'll trust you on it, if the editors want to switch it out. Thanks for the feedback.

      • Lisa Vaas · 733 days ago

        Also, in the same vein, I was tempted to correct Mr. VanHemert on this:

        Yes, Facebook definitely has, as Wired's Kyle VanHemert put it, a mission "to secure its status as the world authority in who knows who..."

        Which of course should be "who knows whom."

        It's a proofreader's goldmine, this story!

      • Who for he, whom for him :)

  4. Keith · 733 days ago

    She should have put General Alexander's face on it.

  5. How exactly does this work and to what end? I want to know the real purpose of such shirts to determine if they can be of good use or not.

  6. how does it work? you give them your money and they give your a shirt that says "im a bit safer and 65 bucks lighter" on it.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

About the author

I've been writing about technology, careers, science and health since 1995. I rose to the lofty heights of Executive Editor for eWEEK, popped out with the 2008 crash, joined the freelancer economy, and am still writing for my beloved peeps at places like Sophos's Naked Security, CIO Mag, ComputerWorld, PC Mag, IT Expert Voice, Software Quality Connection, Time, and the US and British editions of HP's Input/Output. I respond to cash and spicy sites, so don't be shy.