Spammers Solving Difficult AI Problems With An Underground X Prize
from the fascinating dept
Slashdot points us to an interview with Luis von Ahn (who we’re a big fan of), where he talks about how spammers who are frustrated by various types of CAPTCHA tests have set up their own sort of “innovation prize,” offering up somewhere in the range of $500,000 for software that can automatically pass CAPTCHA and reCAPTCHA reading tests (the things where you have to fill in a series of letters to sign up for a service or post a comment). As von Ahn points out: “If [the spammers] are really able to write a programme to read distorted text, great — they have solved an AI problem.” It is, effectively, an “X Prize” for optical character recognition. Not that we like to encourage spammers, but it is rather fascinating how the underground business seems to mirror the above ground innovation world as well.
Filed Under: ai, captcha, luis von ahn, spam, x-prize
Comments on “Spammers Solving Difficult AI Problems With An Underground X Prize”
You know I’d almost pay for software that could effectively do this. I’m not colorblind, my vision is good, I have a good monitor but somehow about 25% of the time when i type them in I get it wrong.
Awesome
For those of you who don’t know, reCAPTCHA is a very popular captcha program that uses images of words that couldn’t be read properly by the OCR software used by various book digitizing programs. The fact that the OCR software failed means that it’s unlikely any other similar software will succeed, which is what makes it such a great captcha.
The neat thing is that it forces you to decipher *two* words, one of which is already known and one which is not. If you get the known word right, you pass. Once enough people give the same answer for a particular unknown word, though, that information is passed back upstream to the book digitizers.
If the spammers can solve this reliably, it means that they’ve made a great advancement in the field of OCR which can be passed back to the book digitizers for great benefits. And it still won’t defeat reCAPTCHA unless the new software is *perfect* – there will still be words that can’t be read by the new software.
Re: Awesome
however, if you were to write a program and seed it with correct answers and have it consistently send the wrong answers back to recaptcha you may be able to trick their servers into accepting wrong answers and then build up your set of solved captchas simply by using a reasonably small set of seeds. … but then again Im not working on the recaptcha so they may already have things in place to prevent this, but there are a ton of sites using it, and i feel like it is almost entirely automated, so this type of hack could potentially ruin their model.
Re: Awesome
Any links to where this is in use?
wait till someone does it, patents it, then gets upset when nonspammers use it and starts suing
DRM effect
Capcha’s may end up having a DRM-like effect as they get harder and harder for humans to to read. They will be keeping out legitimate users, but they may end up allowing easy access to the illicit visitors using CapchaBots.
CAPCHA
I thought CAPCHA had been handily defeated already by offering free porn, and serving up an image capture of the capcha.
—
http://www.chl-tx.com (Thanks, BHO, for the wonderful stimulus you have given *MY* business)
Solved! Gimme $500,000!
http://mturk.amazon.com :-b
We can *totally* trust them to give the award!
No doubt these same honorable folks who manage botnets and such will be men and women of honor. In a worst case scenario, they’ll offer one MILLION dollars in Free V1agra to the winner instead of cash 😛
I’m not sure I’d trust whoever was offering the reward to actually pay me.