CAPTCHA, which stands for "Completely Automated Public Turing test to tell Computers and Humans Apart," is a security tool designed to prevent bots from accessing websites.
A group of scientists in Switzerland optimised an AI model, known as YOLO (You Only Look Once), to master the image-based CAPTCHAs used for identity verification on websites.
Specifically, they focused on reCAPTCHAv2, a version developed by Google that asks users to identify objects like cars, bridges, or traffic lights.
Their research revealed that even if the AI failed on the first attempt, it succeeded in solving a second challenge, and needed only 13 categories of objects to train the model to solve the puzzle.
Further testing demonstrated that the model could outsmart even more advanced CAPTCHAs, which included additional features like mouse movement tracking or browser history analysis.