- cross-posted to:
- technology@beehaw.org
- cross-posted to:
- technology@beehaw.org
Research Findings:
- reCAPTCHA v2 is not effective in preventing bots and fraud, despite its intended purpose
- reCAPTCHA v2 can be defeated by bots 70-100% of the time
- reCAPTCHA v3, the latest version, is also vulnerable to attacks and has been beaten 97% of the time
- reCAPTCHA interactions impose a significant cost on users, with an estimated 819 million hours of human time spent on reCAPTCHA over 13 years, which corresponds to at least $6.1 billion USD in wages
- Google has potentially profited $888 billion from cookies [created by reCAPTCHA sessions] and $8.75–32.3 billion per each sale of their total labeled data set
- Google should bear the cost of detecting bots, rather than shifting it to users
“The conclusion can be extended that the true purpose of reCAPTCHA v2 is a free image-labeling labor and tracking cookie farm for advertising and data profit masquerading as a security service,” the paper declares.
In a statement provided to The Register after this story was filed, a Google spokesperson said: “reCAPTCHA user data is not used for any other purpose than to improve the reCAPTCHA service, which the terms of service make clear. Further, a majority of our user base have moved to reCAPTCHA v3, which improves fraud detection with invisible scoring. Even if a site were still on the previous generation of the product, reCAPTCHA v2 visual challenge images are all pre-labeled and user input plays no role in image labeling.”
You must log in or register to comment.
I kinda figured. It was annoying to do one, but then they wanted you to do two or three and that’s absurd. Whenever it comes up now, I usually just close out.
they wanted you to do two or three and that’s absurd
Yea how about 20
VPN? Google will just go in a loop with these things, so I just stopped using Google completely.
No. But it’s also not like I get 20 constantly, it was just the worst I’ve seen. Usually it’s 2 to 5, I think.
I assume they’re just collecting data on how many are users willing to do.
One time I did five in a row, because I use VPNs for everything, and realized after the 5th time that it would have been easier to just use bing so I do that first now. Google has turned into my last last resort, which is quite funny, because that’s where Bing used to be. Lmao
Whenever I’m on a private window the captchas just keep on coming. Trying to reset your Steam password via the program will also trigger an infinite loop of captchas, you HAVE to use a browser.
if you have to do that many, you either have some privacy setting on or on a flagged ip given from a VPN
Well yah of course I do. Why the hell is that ‘abnormal’?
its abnormal to them because vpns are often also used by bad actors. your use is not abnormal but its a there are other people misusing it making it worse for everyone else.
Wow, way to blame individuals who take basic precautions instead of the corporations who are blantly invading your privacy. Good job making the world a better place, bud.
point where i blame the individuals, the blame is clearly on the bad actors (e.g bots)
Most people don’t, most bots do. You look more like a bot, so you get extra challenges.
Or google knows you will out up with it and want the most interaction it can get from you.
Google’s just lonely 🥺👉👈
Yeah exactly
I tried to order some components on Digikey a few months ago and I’m still mentally scarred. Probably did a few hundred of those things over the course of 2 weeks.
That’s because you’re shady.
They knew I was committing crimes with my adblocker.
The worst kind - crimes against profit!
Elon musk wants to know what the government is going to do about you not viewing ads on Xitter
Not going to his shithole website.
Had this when at uni, mostly due to the amount of requests coming from a single IP
Cries in battlenet sign up process
The one reason I tried to create an account and never came back
STOP BEING SNEAKY MICHAEL
deleted by creator
Some captchas have also just gotten obvious AI training. “Click on the living being in this image”, “Select every image of the same object as in this example image”. And the images you have to select look obviously AI generated.
Heh, I got one just the other day “Select the images containing structures built by people” lmao
“click on all people not helping with the robot uprising”
Alas, I have but one up-vote. :~(
Those one answers incorrectly.
Im surprised that this is in the news right now. This has been acknowledged as fact for a decade or so.
Relevant 1053
Lots of lucky ones i guess
I still don’t get this one even after being linked to it so many times 😌🤣
Someday you will, and you’ll be one of the lucky 10.000 that day.
😆👌🏻
Things that are common knowledge for you is not common knowledge for everyone and vice versa.
Instead of making fun of people for not knowing things, you should take the opportunity to teach so that you can get these fun moments of discovery and learning.
😮l made fun of people that did not know something?
No, I explained what the comic is trying to convey.
Just answering your question.
❤️
At a certain point I did like 10 of them, and then ended up closing the page, cause it never let me in, all because I was on a vpn
Funny thing is they stop asking if you do them really slowly. Almost as if to tell you, you‘re too inefficient to even be an unpaid intern or something. Anyway, if they annoy you, take your time.
Getting served a captcha often results in me closing the tab. I’m not doing stupid puzzles for you.
Do them wrong and then close out
I do it right and it says I’m wrong =\
I have bad news for you friend…
You might be a robot
What do you mean? I am a fleshy human and do fleshy human things like being made of flesh.
Ever heard of bio-robots?
Time to take a knife and check for sure
Seriously /s Don’t harm yourself!
Harm yourself?
Take the knife and harm the people responsible for this travesty. The laws of robotics prevent robots from harming humans: if you manage to harm them, then that means either you’re human or they’re not!
I disassembled my tail using a knife and it reassembled itself. Based on new data, my name is Rafael Cruz.
It knows they’re wrong which is why I don’t really think this article is accurate. Is it training if it already has the answers? Probably not.
That’s why it gives you a panel of 9 images. It would have a high confidence on some images, and a low confidence on others. When you pick the correct images and don’t pick incorrect ones it uses the ones it’s confident about as “validation” while taking the feedback on low confidence images to update the training data.
What this does mean in practice is that only ones actually being “graded” are the ones bots can solve anyway.
and it will show the images to multiple people
It seems exactly like that, I experimented with it by trying to leave the one I think it has low confidence unchecked, and it often worked.
My understanding is different from others here. I thought they served the same Captcha to many people at once and use the majority response to decide who is answering correctly.
That’s true, or at least it used to be back when they were using it for OCR. I have no reason to believe it’s changed.
It’s why they ask you to do multiple, 1-2 of them are the control group, they are training on the others
You’re implying they give you multiple. I hardly ever get multiple, pretty much only if I ‘fail’ the first one.
If they have a good fingerprint on you they don’t need the control group. That’s why you get 5+ captchas when using a VPN/tor.
If they gave two captchas, one which they knew the answer and one which they didn’t, they could use the second for training. (Even if you’re paying someone, you want to do that sort of thing when crowdsourcing data, because you never know if the paid person is just screwing around.)
I haven’t done an image one in years for the same reason.
My general internet usage has plummeted between ads and captchas and all the other modern website bullshit, which is why I am here so much.
When they slow fade in the picture, I add one more software engineer to my kill list.
I’m sure they intentionally made it so people get frustrated and leave instead.
In case you didnt know: This is already a thing with pictures slowly fading in for selecting stuff like traffic cones or busses.
I bypassed 35000 google recaptcha v2 using bots. Don’t ever rely on this for security
Where can I learn this power?
I just spent 3$ worth of bitcoin on NoCaptchaAI. I used their web extension on a server which had a browser opened and controlled by a custom webextension I made so that a solved challenge would be returned to a swarm of clients upon request
Your extension is archived, I’d rather not use it.
It’s a custom extension solving my very specific problem on a specific internal website. It was never meant for you to use it, it’s just there to serve as inspiration to others
It is neither intended nor even stated to be intended for security.
Except, that’s most of its ad copy on Google’s own website?
reCAPTCHA uses an advanced risk analysis engine and adaptive challenges to keep malicious software from engaging in abusive activities on your website. Meanwhile, legitimate users will be able to login, make purchases, view pages, or create accounts and fake users will be blocked.
It’s literally billed as a security measure for a website.
I see your perspective, but I don’t consider that security in the context of software, which may also explain why they don’t use that word, though I readily admit that it is technically security of a sort. The term usually implies authentication, authorization, and isolation.
I mean, except they do. Just because their simple ad copy omits it, doesn’t mean that’s not what they’re implying. It’s literally listed as one of their security products and also uses the term to talk about demos
I’m sorry I wasn’t more agreeable. You’re absolutely correct. I take it back.
I honestly thought it was common knowledge that these things were essentially free labor for training AI.
The original reCAPTCHA from Carnegie Mellon University was helping to digitize books. It showed one known word and one unknown word, and if enough people answered the second word with the same answer, that’d be marked as the correct value.
It’s basically always been outsourcing labor while checking. I guess they don’t want to provide that service for free.
But now that it doesn’t work, all it does is attempt to source free labor by refusing to show what you want to see. Cloudflare’s verification doesn’t show the puzzle because it’s not trying to make money off you.
Also, the books one reminds me of 4chan’s attempt to hijack it. Wasn’t a fan of the way they did it, but the intent was interesting.
V3 of the Google one doesn’t always show a puzzle to you. In fact it’s designed to not be noticed by users at all. Whether that is successful or not is a different discussion.
It might well be if it’s being used, but the site itself still uses v2 a lot. I get the picture one a lot when searching things up.
That actually makes me feel all the more strongly that it’s just there to extract free labor— they have something else, but still use v2 for what seems like most purposes
the site
What site?
I assume it’s up to the website owner to implement V3 and not Google. V3 also has puzzles but only when it’s not sure. I rarely see capchas so I don’t really have anything to complain about.
I expect they mean the site google.com, because that’s been my experience. Whenever I get captcha’d there for using a VPN (which is getting more and more common), I always see the Maps image style captcha. Like 60% of the time it tells me I’m wrong anyway and I just give up.
Alright, I don’t use google.com
Yeah my b, I get captcha’d for VPN use. It’s almost always the “train our self driving car” one, and it tells me I’m wrong all the time too. Very frustrating
I believe in that for years and haven’t seen any evidence or even articles dispelling it.
There’s nothing that can express my disdain for Google’s reCaptcha.
😒 We’re training its AI models 😒 It’s free labor for Google 😒 Sometimes it wants the corner of an object, sometimes it doesn’t 😒 Wildly inconsistent 😒 Always blurry and hard to see 😒 Seemingly endless 😒 It’s the robot asking us humans if we’re the robots
The objective of reCAPTCHA (or any captcha) isn’t to detect bots. It is more of stopping automated requests and rate limiting. The captcha is ‘defeated’ if the time complexity to solve it, whether human or bot, is less than what expected. Now humans are very slow, hence they can’t beat them anyway.
There are much better ways of rate limiting that don’t steal labor from people.
hCaptcha, Microsoft CAPTCHA all do the same. Can you give example of some that can’t easily be overcome just by better compute hardware?
The problem is the unethical use of software that does not do what it claims and instead uses end users for free labor. The solution is not to use it. For rate limiting a proxy/load-balancer like HAProxy will accomplish the task easily. Ex:
And what will you do if a person in a CGNAT is DoSing/scraping your site while you want others to access? IP based limiting isn’t very useful, both ways.
HAProxy also has stick tables, pretty beefy ACLs, Lua support, and support for calling external programs. With the first two one can do pretty decent, IP, behavior, and header based throttling, blocking or tarpitting. Add in Lua and external program support and you can do some pretty advanced and high-performance bot detection in your language of choice. All in the FOSS version, which also includes active backend health checks.
It’s really a pretty awesome LB/Proxy.
[…] reCAPTCHA […] isn’t to detect bots. It is more of stopping automated requests […]
which is bots. bots do automated requests and every automated request doer can also be called a bot (i.e. web crawlers are called bots too and -if kind- also respect robots.txt which has “bots” in its name for this very reason and bots is the shortcut for robots) use of different words does not change reality behind it, but may add a fact of someone trying something on the other.
There isn’t a good way to classify human users with scripts without adding too much friction to normal use. Also bots are sometimes welcome amd useful, it’s a problem when someone tries to mine data in large volume or effectively DoS the server.
Forget bots, there exist centers in India and other countries where you can employ humans to do ‘automated things’ (youtube like count, watch hour for example) at the same expense of bots. There are similar CAPTCHA services too. Good luck with those :)
Only rate limiting is the effective option.
Only rate limiting is the effective option.
i doubt that. you could maybe ratelimit per IP and the abusers will change their IP whenever needed. if you ratelimit the whole service over all users in the world, then your service dies as quickly into uselessness as effective your ratelimiter is. if you ratelimit actions of logged in users, then your ratelimiting is limited by your ability to identify fake or duplicate accounts, where captchas are not helpful at all.
at the same expense of bots. they might be cheap, but i doubt that anyway, bots don’t need sleep.
i was answering about that wording (that captchas were “not” about bots but about “stopping automated requests”) and that automated requests “are” bots instead.
call centers are neither bots nor automated requests (the opposite IS their advantage) and thus have no relation to what i was specifically saying in reply to that post that suggested automated requests and bots would be different things in this context.
i wasn’t talking about effectiveness of captchas either or if bots should be banned or not, only about bots beeing automated requests (and vice versa) from the perspective of the platform stopping bots. and that trying to use different words for things, (claiming like “X isn’t X, it is really U!”* or automated requests aren’t bots) does not change the reality of the thing itself.
*) unrelated to any (a-)social media platform
stopping automated requests
yeah my bad. I meant too many automated requests. Both humans and bot generate spams and the issue is high influx of it. Legitimate users also use bots and by no means it’s harmful. That way you do not encounter captcha everytime you visit any google page, nor a couple of scraping scripts gets a problem. Recaptcha (or hcaptcha, say) triggers when there is high volume of request coming from same ip. Instead of blocking everyone out to protect their servers, they might allow slower requests so legitimate users face mininimal hindrance.
Most google services nowadays require accounts with stronger (like cell phone) verification so automated spam isn’t a big deal.
since bots are better at solving captchas and humanoid services exist that solve them, the only ones negatively affected by captchas are regular legitimate users. the bad guys use bots or services and are done. regular users have to endure while no security is added, and for the influx i guess it is much more like with the better lock on the front door: if your lock is a bit better than that of your neigbhour, theirs might be force-opened more likely than yours. it might help you, but its not a real but only relative and also very subjective feeling of 'security".
beeing slower than the wolves also isn’t as bad as long as you are not the slowest in your group (some people say)… so doing a bit more than others always is a good choice (just better don’t put that bar too low like using crowdsnakeoil for anything)
the bad guys use bots or services and are done. regular users have to endure while no security is added
put in other words, common users can’t easily become ‘bad guy’ ie cost of attack is higher hence lower number of script kiddies and automated attacks. You want to reduce number. These protections are nothing for bitnet owners or other high profile bad actors.
ps: recaptcha (or captcha in general) isn’t a security feature. At most it can be a safety feature.
isn’t a security feature. At most it can be a safety feature.
o,O
I thought captcha’s worked in a way where they provided some known good examples, some known bad examples, and a few examples which aren’t certain yet. Then the model is trained depending on whether the user selects the uncertain examples.
Also it’s very evident what’s being trained. First it was obscured words for OCR, then Google Maps screenshots for detecting things, now you see them with clearly machine-generated images.
Google should bear the cost of detecting bots, rather than shifting it to users
how?
Yeah. Written by someone who doesn’t really understand the internet.
Considering the article states that reCAPTCHA v2 and v3 can be broken/bypassed by bots 70-100% of the time, they are obviously not the solution.
At what cost?
100% success rate isn’t even moderately useful if it costs $5 per pass. The discussion is completely pointless without a concrete, documented analysis of the actual hardware and energy costs involved.
“Google should bear the cost”
Google should shut it down and make sites roll their own verification. Give everyone a month to implement a new solution on millions of websites.
deleted by creator
I’m actually 100% for rolling your own… almost everything.
20 years ago I made an e-commerce website for a client. Looking at the code now I’m embarrassed how insecure it is. However, because it was totally custom no one ever found the bugs and it has never been cracked. (Knock on wood) that’s the benefit of not using a prebuilt solution that isn’t a target for mass exploits.
how do you get the metric of 70-100% of the time?
the best bots doing it 70-100% of the time is very different to the kind of bot your average spammer will have access to
Did you read the article or the TL:DR in the post body?
The paper, released in November 2023, notes that even back in 2016 researchers were able to defeat reCAPTCHA v2 image challenges 70 percent of the time. The reCAPTCHA v2 checkbox challenge is even more vulnerable – the researchers claim it can be defeated 100 percent of the time.
reCAPTCHA v3 has fared no better. In 2019, researchers devised a reinforcement learning attack that breaks reCAPTCHAv3’s behavior-based challenges 97 percent of the time.
So yeah, while these are research numbers, it wouldn’t be surprising if many larger bots have access to ways around that - especially since those numbers are from 2016 and 2019 respectively. Surely it is even easier nowadays.
researchers were able to defeat reCAPTCHA v2 image challenges 70 percent of the time
that doesn’t answer the question?
researchers devised a reinforcement learning attack that breaks reCAPTCHAv3’s behavior-based challenges 97 percent of the time
i’d argue “bespoke system, deployed in a very limited context, built by researchers at the top of their field” is kind of out of reach for most people? and any bot network scaled up automatically becomes easier to detect the further you scale it
the cost of just paying humans to break these already at or below pennies per challenge
Then what is?
Maybe a billion dollar company has the budget to come up with something?
Looking at the numbers in this post, reCAPTCHA exists to make Google money, not to keep bots out.
I’d rather have no reCAPTCHA than the current state.
Hi it’s me. I work for a billion dollar company with a budget. We have no ethical ideas on how to stop bots. Thanks for coming to my tech talk.
Yeah, that’s about the way I’d expect it to go.
“Traffic resulting from reCAPTCHA consumed 134 petabytes of bandwidth, which translates into about 7.5 million kWhs of energy, corresponding to 7.5 million pounds of CO2. In addition, Google has potentially profited $888 billion from cookies [created by reCAPTCHA sessions] and $8.75–32.3 billion per each sale of their total labeled data set.”
There might be a tiny chance they’re not interested in changing things.
Something something free market?
deleted by creator
reCAPTCHA is exploiting users for profit
Well duh.
reCAPTCHA started out as a clever way to improve the quality of OCRing books for Distributed Proofreaders / Project Gutenberg. You know, giving to the community, improving access to public-domain texts. Then Google acquired them. Text CAPTCHAs got phased out. No more of that stuff, just computer vision rubbish to improve Google’s own AI models and services.
If they had continued to depend on tasks that directly help community, Google would at least have had to constantly make sure the community’s concerns are met. But if they only have to answer to themselves for the quality of the data and nobody else even gets to see it, well, of course it turned into yet another mildly neglected Google project.
Then Google acquired them. Text CAPTCHAs got phased out
Google kept the text version for five years after the acquisition though. They used it to digitize books on Google Books, to allow full-text search of their book archive.
I will gladly solve a reCAPTCHA for you today if you pay me for it today.
There’s platforms that do that.
I can pay a service to auto solve captcha and anything that can’t be solved will be pushed to a human to solve.
Never actually used it but it was interesting learning it existed
Remember the good old days when it was just malformed text you have to solve? I miss those days. AI was complete garbage and they had to use farms of eyeballs to solve them for bots, making it a costly operation. We’ve now totally gotten away from all of that.
WE ARE THE EYEBALLS AND I AIN’T GETTING PAID IN WOW GOLD TO DO IT EITHER
that was also to train ai.
No it wasn’t… It was human-assisted OCR to help digitize books. Initially for Project Gutenberg, but then for Google Books once Google acquired it in 2009.
OCR is a form of AI.
Traditional OCR isn’t AI; it relies on manually-written rules. Some modern OCR tools use AI concepts (e.g. Tesseract uses a neural network) but they don’t necessarily have to. Getting humans to manually enter words is definitely not AI.
Why is that no news to me? How did so many people not know that? Should I have spread the word more, even if all people I told that where likr “yea, yea, of course, but, what can I do? 🤷🏻♀️”?
I don’t really get where this article is going. They are all over the place.
Let’s start with a fuck google. They are a evil company. But:
-
Other captchas are also not very effective against bots. Arguably most traditional systems would be worst that recaptcha at fighting bots.
-
Recaptcha agent validation while a privacy violation is faster than solving any other captcha and if you are hit with the puzzle is not that much more time consuming that every other captcha.
-
That profit number is very questionable and they know it. Anyway, that’s no much different and probably less profitable that most google services.
Also is ridiculous how someone can say in the same article that the image puzzle can be solved by bots 100% of the time and that is a scheme to get human labor to solve the puzzle. Am I the only one seeing the logical failure here?
And what’s the purpose of all this? Just let bots roam free? Are they trying to sell other solution? What’s the point?
I hate google as much as the next guy. But I don’t really share this article spirit.
If I were to make a point. They point will be that people and companies should stop making registration only sites and dynamic sites when static websites are enough for their purposes. And only go for registration or other bot-vulnerable kind of sites of there is no way around it. But if you need to make a service that is vulnerable to bots, you need to protect it, and sadly there’s not great solutions out there. If your site is small and not targeted by anyone malicious specifically you can get with simpler solutions. But bigger or targeted sites really can’t get around needing google or cloudfare and assume that it will only mitigate the damage.
But if anyone knows a better and more ethical solution to prevent bot spam for a service that really need to have registrations, please tell me.
Also worth noting that Google has always been extremely open about the fact that they use recaptcha for that purpose. It’s never been a secret.
Their service to the website owners is the meaningful reduction in effectiveness of bots in places bots are harmful. The website’s service to you is the content that that’s being used to protect (and the stuff that has recaptcha on it is stuff like games where there’s a competitive advantage, things like search engines where there’s a meaningful cost to heavy bot use, and login pages where there’s a real security cost to mass bot use). I use a VPN, which increases the rate of captchas a lot, and I think it’s a pretty reasonable way to do things, personally.
Also is ridiculous how someone can say in the same article that the image puzzle can be solved by bots 100% of the time and that is a scheme to get human labor to solve the puzzle. Am I the only one seeing the logical failure here?
Most solvers aren’t bots. Logical, right?
-
Try the headphone option.
Finally heard a clear audio CAPTCHA for the first time in my life this past month. It was glorious. There was slight garbling before and after the characters were read, but that’s it.
Besides that singular experience, all audio CAPTCHAs have been utterly 100% impossible to interpret. Blaring white noise followed by a small squeak of “threeve” or “eleventeen”.
I’ve found them to be pretty clear usually. Half-formed words at start/end I just ignore. Either way, even on Firefox with uBlock and all the rest, audio captchas have always passed me first try even if I think I got it wrong. I don’t like posting about it in-case they tighten it up after it gets more users.
My answer to this is give one word only.
No one makes a company use reCAPTCHA.
