Submitted via IRC for takyon
Talk about a GAN-do attitude... AI software bots can see through your text CAPTCHAs
[...] Boffins at Lancaster University in the UK, Northwest University in the US, and Peking University in China have devised an approach for creating text-based CAPTCHA solvers that makes it trivial to automatically decipher scrambled depictions of text.
Researchers Guixin Ye, Zhanyong Tang, Dingyi Fang, Zhanxing Zhu, Yansong Feng, Pengfei Xu, Xiaojiang Chen, and Zheng Wang describe their CAPTCHA cracking system in a paper that was presented at the 25th ACM Conference on Computer and Communications Security in October and now released to the public.
As can be surmised from the title, "Yet Another Text Captcha Solver: A Generative Adversarial Network Based Approach," the computer scientists used a GAN (Generative Adversarial Network) to teach their CAPTCHA generator, which is used for training their text recognition model.
First described in 2014, a GAN consists of two neural network models pitted against each other as adversaries, one simulating something and the other spotting problems with the simulation until any differences can not longer be identified.
Coincidentally, that's the same year researchers from Google and Stanford published a paper titled, "The End is Nigh: Generic Solving of Text-based CAPTCHAs." Four years on, the speed bumps limiting generic attacks have been paved over.
A GAN turns out to be well-suited for efficiently training data models. It allowed the researchers to teach their CAPTCHA generation program to quickly create lots of synthetic text puzzles to train their basic puzzle solving model. They then fine-tuned it via transfer learning to defeat real text jumbles using only a small set (~500 instead of millions) of actual samples.
(Score: 3, Insightful) by bzipitidoo on Friday December 07 2018, @02:17PM (4 children)
For captchas, just use scanned text. Optical Character Recognition can never get those right. Burn -> bum, skill -> skin, kettle-> ke#1e.
Or maybe we could use this AI software for some decent OCR?
(Score: 5, Informative) by Anonymous Coward on Friday December 07 2018, @02:35PM (2 children)
Captchas exist to force people to turn on scripting, not to keep bots out.
(Score: 2) by crafoo on Friday December 07 2018, @11:23PM (1 child)
Yep. was going to mark funny, but I think this is probably true. It's also good to collect data to measure how good on average a human is at solving captchas. This is useful in determining how good your NN is at solving them - better or worse than human.
(Score: 1) by Ethanol-fueled on Saturday December 08 2018, @02:37AM
No, it is true. And besides scripting, if you wanted to post something anonymously via ToR, for example, you can't, because any forum armed with Jewgle's captchas only returns you Jewgle's "Hahah, fuck you buddy, denied" message*. Oh well, it was fun while it lasted. And some snarky messages are be left not being sent rather than wardriving, cracking networks, or sitting outside Starbucks with a laptop and a ski-mask.
* We "think you may be a robot," or something.
(Score: 0) by Anonymous Coward on Friday December 07 2018, @03:18PM
Google originally did use their scanned text for recaptcha, so captcha solvers would provide feedback to train the Google OCR. Now they train autonomous cars instead.
On the other side, captchas are a bit different from OCR on a scan: when you input a candidate solution you get immediate feedback whether or not the solution is correct. If the solution is incorrect, you get to try again, often as many times as necessary. So a bot doesn't need to get it right every time; even if the bot can solve it even 1% of the time that's probably entirely good enough for a captcha solver. While if you are doing OCR on a scanned document... if only 1% of the words were correct then that's probably worse than useless.
(Score: 3, Funny) by All Your Lawn Are Belong To Us on Friday December 07 2018, @03:15PM (4 children)
ReCAPTCHA doesn't, anyway. But I guess Pepperidge Farm remembers.....
This sig for rent.
(Score: 3, Insightful) by edIII on Friday December 07 2018, @08:59PM (3 children)
I dunno, but who the fuck can pass one these days? I've sat there for over a minute clicking fucking tile after tile after tile. I've outright failed to get through some captcha systems. Meanwhile the bots keep evolving with AI, and can solve the captchas faster than any human can.
At some point proving your human is going to mean including random failures in specific ways, or fucking up in a way only a human can do :)
Server: Are you human?
Bot: Ummm? Derpity Derp Derp! Fuck U. I need to login fuk. #bullshit #derp
Server: You may pass
Technically, lunchtime is at any moment. It's just a wave function.
(Score: 1, Insightful) by Anonymous Coward on Friday December 07 2018, @10:52PM (2 children)
Using you as a mechanical turk.
Also likely using it to help them identify your mouse/click profile for future identification on other sites. I've been having to take a few minutes each time on those captchas making my mouse movements and clicks look fake, and having both clicked wrong tiles and clicked exactly the correct tiles each time, they always take a minimum of 3 and up to 10 windows worth of times before accepting you as human and letting you in.
It is free labor, not you fucking up as a human that is making the google captcha experience so bad. Do what I do and unless you HAVE TO use a site, just say No! to google.
(Score: 3, Interesting) by edIII on Saturday December 08 2018, @01:27AM (1 child)
Ohhh, I've gone much, much, much, farther than that. There have been some big name vendor sites that started using Google's recaptcha at every single login. I totally lost my shit one day during a service outage when the captcha system was preventing me from accessing the administrative functions I desperately needed. Called up technical support and I made them make the changes for me. Not initially, but after two levels of supervisors they did.
After that, their CTO was surprised that I got all the way to him, and I read him the fucking riot act. I demanded as a consequence of my obvious stupidity, and received, an IP white listing from a few management networks that prevented the captcha from ever appearing again. We agreed too, that the website itself was a fallback measure, and I should encode all the adminstrative functions into my platform via an API. I agreed, but it is still some work needing to be done to fully integrate it.
All because a fucking captcha system is too damn complicated for a human to get past. Fuck if know why I kept failing. I clicked all the goddamn fucking boxes with a car in it, or a sign it, or a storefront it. If they were going to torture my ass, the least they could do is create a pornographic one. Pick all the DPs in the following picture....
Technically, lunchtime is at any moment. It's just a wave function.
(Score: 0) by Anonymous Coward on Saturday December 08 2018, @09:15PM
"Fuck if know why I kept failing."
Uh oh, won't be long before the bots suspect they might not be human. This one is just 3rd gen, I'm betting the sixes are already aware.
(Score: 1) by progo on Friday December 07 2018, @03:21PM (6 children)
Does BBC use the word 'boffin'? The local colloquialism annoys me.
It's like the NY Post can't mention anything having to do with the New York Subway without mentioning 'straphangers'. Literally no one in NYC says 'straphangers' except the Post. I think they think they're cool or something for using the word.
(Score: 3, Insightful) by takyon on Friday December 07 2018, @03:38PM
Well, your vocab has been successfully expanded but the latent triggering still needs to be removed.
YOLO.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2, Funny) by Anonymous Coward on Friday December 07 2018, @04:48PM (4 children)
The British like to play with language a lot. It distracts them from the fact that they are citizens of a tiny nation isolated on a tiny island. Plus they habitually drink warm, muddy beer, and this destroys their palettes, leading them to consume really, really bad food.
So forgive them, and Robert's your mother's brother, eh?
(Score: 0) by Anonymous Coward on Friday December 07 2018, @11:47PM (3 children)
So says the AC from the land of hamburgers, hotdogs and twinkies.
(Score: 0) by Anonymous Coward on Saturday December 08 2018, @07:06PM (2 children)
UK: Patty on a bap. Very common food there. Hamburgers, like most sandwich stacks, can range from pretty boring to utterly sublime. At least here in the US. Sorry your taste buds were torn off in the football* stadium riots.
UK: Sausage. Very common food there. Yours just aren't as good as ours. A shame.
Twinkies are available in the UK and Ireland under the Hostess brand name where they're sold in Sainsburys, Tesco, ASDA and B&M stores. Very common snack food there. Although frankly, Twinkies were a lot better when they were fresh-baked products not infused with so many preservatives.
So...
Yeah. You got nothing. See, I've lived in both places, and American food is way better. :)
So head on back out to the pub, suck down some warm, cloudy beer, and enjoy your choice of haggis, stargazy pie, black pudding, and offal. Perhaps a side of some drippings on toast. Throw some liquor** on the latter for extra... flavo(u)r.
Notes for US readers:
* "football" is what they call soccer in the UK
** "liquor" in the UK is a (horrible-tasting) sauce made with parsley and vinegar. Seriously.
(Score: 1, Informative) by Anonymous Coward on Saturday December 08 2018, @07:54PM (1 child)
UK reader here.
Yes, because we invented it, and because it involves moving a ball with your foot ... unlike the so-called American football, in which the ball is very rarely kicked but handled most of the time.
Never heard of it. Liquor = alcoholic spirit, as in the US, but it's rarely used. Instead of liquor stores, we have off-licences, the meaning of which should be self-evident. ;-)
Again, never heard of them. I don't deny they may exist, but I've never seen them or consumed one.
It's not warm, it's served at room temperature. And if it's cloudy, you send it back; it should be clear.
I can't be bothered to comment on the rest, but it's mostly rubbish (=garbage, trash, crap).
AC
(Score: 2) by takyon on Monday December 10 2018, @03:52AM
Not all beer should be clear. Where did you or the other AC even get that idea?
Bunch of examples listed here:
https://beerconnoisseur.com/articles/what-unfiltered-beer [beerconnoisseur.com]
IPAs, wheat beers, stouts, etc. can all be hazy and taste great.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 5, Funny) by Freeman on Friday December 07 2018, @04:45PM (2 children)
Soon to be an obsolete method of detection: https://xkcd.com/632/ [xkcd.com]
Joshua 1:9 "Be strong and of a good courage; be not afraid, neither be thou dismayed: for the Lord thy God is with thee"
(Score: 2, Funny) by nitehawk214 on Friday December 07 2018, @06:31PM (1 child)
Also appropriate: https://xkcd.com/810/ [xkcd.com]
"Don't you ever miss the days when you used to be nostalgic?" -Loiosh
(Score: 0) by Anonymous Coward on Saturday December 08 2018, @09:17PM
Real chuckles from that one thanks! The OP was good too.