> Our work examines the efficacy of employing advanced machine learning methods to solve captchas from Google's reCAPTCHAv2 system. We evaluate the effectiveness of automated systems in solving captchas by utilizing advanced YOLO models for image segmentation and classification. Our main result is that we can solve 100% of the captchas, while previous work only solved 68-71%. Furthermore, our findings suggest that there is no significant difference in the number of challenges humans and bots must solve to pass the captchas in reCAPTCHAv2. This implies that current AI technologies can exploit advanced image-based captchas. We also look under the hood of reCAPTCHAv2, and find evidence that reCAPTCHAv2 is heavily based on cookie and browser history data when evaluating whether a user is human or not. The code is provided alongside this paper.
> CAPTCHAs (Completely Automated Public Turing Tests to Tell Computers and Humans Apart) have been a vital security measure on the internet, protecting websites from automated bots and malicious activities. However, with the rapid advancements in machine learning and artificial intelligence, the effectiveness of CAPTCHAs in distinguishing between humans and machines has come into question.
> This semester project focuses on Google's reCAPTCHAv2 system, which is widely used across the web. We aim to analyze the effectiveness of reCAPTCHAv2 in rejecting bots using advanced deep learning models such as YOLO (You Only Look Once). Our research explores the vulnerabilities of image-based CAPTCHAs and develops efficient methods to solve them using state-of-the-art machine learning techniques.
> Our main findings include:
> We can solve 100% of the CAPTCHAs presented by reCAPTCHAv2, surpassing the success rates of previous works, which range from 68% to 71%.
> There is no significant difference in the number of challenges required by humans and bots to solve CAPTCHAs in reCAPTCHAv2, suggesting that current AI technologies can effectively exploit advanced image-based CAPTCHAs.
> reCAPTCHAv2 heavily relies on cookie and browser history data when evaluating whether a user is human or not, highlighting the importance of considering user-specific data in CAPTCHA design.
2 comments
[ 3.4 ms ] story [ 16.2 ms ] thread