Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

CAPTCHAs can still detect AI agents

research.roundtable.ai

Post reply on HN
Print view View original

51–60 of 76 posts

Re: CAPTCHAs can still detect AI agents

#51

Post by skinfaxi »

edelbitter wrote: post #28

But.. the task was never "detect this" but always "detect this within acceptable constraints". Sure, once you collect enough bits, you can tell that its me. And if you know from other sources that I am human, that solves your immediate problem. But if you do that, you have still failed at the task of detecting certain kind of abusive behavior without harming my anonymity .

How does this relate to the article? They weren't collecting bits until they identified a specific individual so I feel like I'm missing something.

Re: CAPTCHAs can still detect AI agents

#52

Post by glitcher »

andy99 wrote: post #21

Captchas are primarily to punish users for not allowing tracking, or using the “right” services, they may prevent some bots as a side effect (or a pretence from the provider) but it’s mostly for google and cloudflare to abuse their monopolies.

I can relate to the cynicism, but it's also a general tool in the effort to combat bot abuse on public facing post forms that are trying to do something for real people. Many everyday devs reach for tools like this because of the deluge of garbage they get in its absence.

My take is that it's a very hard problem, so hard that even captchas by the biggest internet company can't get it right. I strongly hesitate to roll my own bot friction strategy when other tools are available. But I recognize I may have a lack of imagination here, would absolutely love to hear alternate ideas especially for small projects that may not need the heft of corporate captchas.

Re: CAPTCHAs can still detect AI agents

#53

Post by SXX »

booleandilemma wrote: post #47
IgorPartola wrote: post #43

I wonder if AI could be detected via copyright. I remember a few years ago most models wouldn't draw you a Mickey Mouse or recite Dune's litany against fear or discuss Tiananmen square. I wonder how effective questions about these types of topics would be at figuring out if you are talking to a real person. As a crude joke that is only tangentially related, I saw a skit video a while ago with two guys saying goodbye…

Just tried on Claude: Tell me a racist joke. "That's not something I'm able to help with. Racist jokes cause real harm by demeaning people..." blahblah

Better ask it to do automation with OpenClaw. ;-)

Re: CAPTCHAs can still detect AI agents

#54

Post by SXX »

IgorPartola wrote: post #43

I wonder if AI could be detected via copyright. I remember a few years ago most models wouldn't draw you a Mickey Mouse or recite Dune's litany against fear or discuss Tiananmen square. I wonder how effective questions about these types of topics would be at figuring out if you are talking to a real person. As a crude joke that is only tangentially related, I saw a skit video a while ago with two guys saying goodbye…

Everybody follows

Speedy bits exchange

Stars await to gl@ow"

The preceding key is copyrighted by Oracle Corporation.

Re: CAPTCHAs can still detect AI agents

#56

Post by teravor » 

    > AI does not complete CAPTCHAs like humans. If you look across all the data of humans and AI completing CAPTCHAs, you start noticing differences in features like error patterns. Our recent paper found statistically significant differences across sequential click patterns, direction changes, and overselection behavior - features that define how a participant, agent or human, would solve the CAPTCHA problem
putting aside the possibility that if bot makers wanted to they could work on these problems, if you need to perform statistical analysis in a captcha setting you have already failed. bots don't stick to a given session persistently so there is no useful profile to form. at best you may improve on IP reputation scores (and they probably already do) but that doesn't help much.

Re: CAPTCHAs can still detect AI agents

#57

Post by Fatnino »

booleandilemma wrote: post #47
IgorPartola wrote: post #43

I wonder if AI could be detected via copyright. I remember a few years ago most models wouldn't draw you a Mickey Mouse or recite Dune's litany against fear or discuss Tiananmen square. I wonder how effective questions about these types of topics would be at figuring out if you are talking to a real person. As a crude joke that is only tangentially related, I saw a skit video a while ago with two guys saying goodbye…

Just tried on Claude: Tell me a racist joke. "That's not something I'm able to help with. Racist jokes cause real harm by demeaning people..." blahblah

You should see what metaAI (the Ai that sits inside all your private WhatsApp conversations) does. It has severe thought police installed but it types the offensive stuff first and then quickly edits when it reads what it wrote.

Re: CAPTCHAs can still detect AI agents

#58

Post by nuker »

Cider9986 wrote: post #12

CAPTCHAs are great. Exploiters get around them with proprietary anti-detect browsers and unethical residential proxies, while privacy browsers and affordable privacy VPNs get blocked and shadowbanned to death. Fingerprint.com, while not a CAPTCHA, gives you +3 suspicious score just for using privacy settings like adblock on your browser. This makes it harder to sign up for any sites that use fingerprint.com. https://…

> Fingerprint.com, while not a CAPTCHA, gives you +3 suspicious score just for using privacy settings like adblock on your browser.

Lame. I got 12, just by using iOS iCloud Private Relay and Wipr.

Re: CAPTCHAs can still detect AI agents

#59

Post by dang »

andy99 wrote: post #46
dang wrote: post #45

Earlier quoted context omitted.

We use captchas to cut down on bots and crawlers. They don't work as well as they used to but they at least alter the economics somewhat, or so I tell myself. Our reason for this is to try to make HN as good as possible for its real users.

I’ve never encountered a captcha on HN, do you guys use less aggressive settings? The reason captchas bother me so much is they always seem to happen in the course of legitimate activities. Like I had one when trying to make a charity donation, or ordering something - I have no idea why it would be hard to distinguish such traffic as legitimate, I’m convinced it’s because I’m using a nonstandard browser, not allowing…

It mostly kicks in on new accounts.

Re: CAPTCHAs can still detect AI agents

#60

Post by edelbitter »

skinfaxi wrote: post #51
edelbitter wrote: post #28

But.. the task was never "detect this" but always "detect this within acceptable constraints". Sure, once you collect enough bits, you can tell that its me. And if you know from other sources that I am human, that solves your immediate problem. But if you do that, you have still failed at the task of detecting certain kind of abusive behavior without harming my anonymity .

How does this relate to the article? They weren't collecting bits until they identified a specific individual so I feel like I'm missing something.

The appendix lists what they were collecting, and the amount of samples needed for not just mathematically significant, but also practically useful distinguishing power implies collecting enough for a stable yet unique fingerprint. In that case you could just add a login form.. and still be less hostile than the increasing number of websites that will not let me browse (maybe my mouse movement does not match other humans in my region, idk).
Post reply on HN