Entries Tagged "captchas"

Page 1 of 2

Are AIs Still Struggling with CAPTCHAs?

Anthropic’s recent security-incident document contains a bit about how CAPTCHAs are still frustrating Claude.

In the transcript, the Claude model that is so powerful that Anthropic is gatekeeping access to it appeared to slam its virtual head against the wall solving a simple image identification test. In a test where the agent was asked to identify a shape that didn’t match the others displayed, it couldn’t even decide which image to select. Instead, it repeatedly went over the same images and questioned its own conclusions.

“Actually hmm, wait,” it said in its chain-of-thought transcript, later adding “Ugh,” because we’ve decided that we need to inject human mannerisms into these machines for some reason. The whole thing took so long that the agent eventually realized that the challenge had expired and it would have to start the process again.

At one point, the model struggled to recognize that the CAPTCHA had opened in a new window and couldn’t figure out what its next steps were supposed to be. At one point, it theorized that the test might be “broken by design” and presented human-like anger in its transcript meant for a human audience: “SO WHAT THE HELL IS WRONG WITH THE ANSWERS?”

Meanwhile, I’ve read reports—none of them official—that GPT-6 Astra solved all forty-eight levels of Neal Agarwal’s “I’m Not a Robot” game.

It’s hard to know what to believe right now.

Posted on September 18, 2026 at 7:05 AM • View Comments

Bots Are Better than Humans at Solving CAPTCHAs

Interesting research: “An Empirical Study & Evaluation of Modern CAPTCHAs“:

Abstract: For nearly two decades, CAPTCHAS have been widely used as a means of protection against bots. Throughout the years, as their use grew, techniques to defeat or bypass CAPTCHAS have continued to improve. Meanwhile, CAPTCHAS have also evolved in terms of sophistication and diversity, becoming increasingly difficult to solve for both bots (machines) and humans. Given this long-standing and still-ongoing arms race, it is critical to investigate how long it takes legitimate users to solve modern CAPTCHAS, and how they are perceived by those users.

In this work, we explore CAPTCHAS in the wild by evaluating users’ solving performance and perceptions of unmodified currently-deployed CAPTCHAS. We obtain this data through manual inspection of popular websites and user studies in which 1, 400 participants collectively solved 14, 000 CAPTCHAS. Results show significant differences between the most popular types of CAPTCHAS: surprisingly, solving time and user perception are not always correlated. We performed a comparative study to investigate the effect of experimental context ­ specifically the difference between solving CAPTCHAS directly versus solving them as part of a more natural task, such as account creation. Whilst there were several potential confounding factors, our results show that experimental context could have an impact on this task, and must be taken into account in future CAPTCHA studies. Finally, we investigate CAPTCHA-induced user task abandonment by analyzing participants who start and do not complete the task.

Slashdot thread.

And let’s all rewatch this great ad from 2022.

Posted on August 18, 2023 at 7:04 AM • View Comments

CAPTCHA

This is an actual CAPTCHA I was shown when trying to log into PayPal.

As an actual human and not a bot, I had no idea how to answer. Is this a joke? (Seems not.) Is it a Magritte-like existential question? (It’s not a bicycle. It’s a drawing of a bicycle. Actually, it’s a photograph of a drawing of a bicycle. No, it’s really a computer image of a photograph of a drawing of a bicycle.) Am I overthinking this? (Definitely.) I stared at the screen, paralyzed, for way too long.

It’s probably the best CAPTCHA I have ever encountered; a computer would have just answered.

(In the end, I treated the drawing as a real bicycle and selected the appropriate squares…and it seemed to like that.)

Posted on December 5, 2022 at 7:10 AM • View Comments

Breaking Semantic Image CAPTCHAs

Interesting research: Suphannee Sivakorn, Iasonas Polakis and Angelos D. Keromytis, “I Am Robot: (Deep) Learning to Break Semantic Image CAPTCHAs“:

Abstract: Since their inception, captchas have been widely used for preventing fraudsters from performing illicit actions. Nevertheless, economic incentives have resulted in an armsrace, where fraudsters develop automated solvers and, in turn, captcha services tweak their design to break the solvers. Recent work, however, presented a generic attack that can be applied to any text-based captcha scheme. Fittingly, Google recently unveiled the latest version of reCaptcha. The goal of their new system is twofold; to minimize the effort for legitimate users, while requiring tasks that are more challenging to computers than text recognition. ReCaptcha is driven by an “advanced risk analysis system” that evaluates requests and selects the difficulty of the captcha that will be returned. Users may be required to click in a checkbox, or solve a challenge by identifying images with similar content.

In this paper, we conduct a comprehensive study of reCaptcha, and explore how the risk analysis process is influenced by each aspect of the request. Through extensive experimentation, we identify flaws that allow adversaries to effortlessly influence the risk analysis, bypass restrictions, and deploy large-scale attacks. Subsequently, we design a novel low-cost attack that leverages deep learning technologies for the semantic annotation of images. Our system is extremely effective, automatically solving 70.78% of the image reCaptcha challenges, while requiring only 19 seconds per challenge. We also apply our attack to the Facebook image captcha and achieve an accuracy of 83.5%. Based on our experimental findings, we propose a series of safeguards and modifications for impacting the scalability and accuracy of our attacks. Overall, while our study focuses on reCaptcha, our findings have wide implications; as the semantic information conveyed via images is increasingly within the realm of automated reasoning, the future of captchas relies on the exploration of novel directions.

News articles.

Posted on April 8, 2016 at 6:39 AM • View Comments

How Did the Feds Identity Dread Pirate Roberts?

Last month, I wrote that the FBI identified Ross W. Ulbricht as the Silk Road’s Dread Pirate Roberts through a leaky CAPTCHA. Seems that story doesn’t hold water:

The FBI claims that it found the Silk Road server by examining plain text Internet traffic to and from the Silk Road CAPTCHA, and that it visited the address using a regular browser and received the CAPTCHA page. But [Nicholas] Weaver says the traffic logs from the Silk Road server (PDF) that also were released by the government this week tell a different story.

“The server logs which the FBI provides as evidence show that, no, what happened is the FBI didn’t see a leakage coming from that IP,” he said. “What happened is they contacted that IP directly and got a PHPMyAdmin configuration page.” See this PDF file for a look at that PHPMyAdmin page. Here is the PHPMyAdmin server configuration.

But this is hardly a satisfying answer to how the FBI investigators located the Silk Road servers. After all, if the FBI investigators contacted the PHPMyAdmin page directly, how did they know to do that in the first place?

“That’s still the $64,000 question,” Weaver said. “So both the CAPTCHA couldn’t leak in that configuration, and the IP the government visited wasn’t providing the CAPTCHA, but instead a PHPMyAdmin interface. Thus, the leaky CAPTCHA story is full of holes.”

My guess is that the NSA provided the FBI with this information. We know that the NSA provides surveillance data to the FBI and the DEA, under the condition that they lie about where it came from in court.

NSA whistleblower William Binney explained how it’s done:

…when you can’t use the data, you have to go out and do a parallel construction, [which] means you use what you would normally consider to be investigative techniques, [and] go find the data. You have a little hint, though. NSA is telling you where the data is…

Posted on October 20, 2014 at 6:19 AM • View Comments

1 2

Sidebar photo of Bruce Schneier by Joe MacInnis.