Google’s popular reCAPTCHA system, designed to distinguish between humans and bots on online website is now cracked by AI.
It has been completely cracked by researchers at ETH Zurich. Using advanced image recognition algorithms, the team achieved a 100% success rate in bypassing all types of reCAPTCHAv2 challenges.
The researchers used YOLO models for image segmentation and classification, training them on a dataset of 14,000 labeled images.
They discovered that reCAPTCHAv2 heavily relies on cookie and browser data to identify human users. By combining their image recognition system with a VPN, realistic mouse movements, and browser data, the team’s automated solution went completely undetected.
This development raises serious questions about the effectiveness of current CAPTCHA systems in protecting websites from automated access.
The ETH Zurich team has made their source code public to encourage further research in this area.
As online security continues to evolve, this breakthrough serves as a wake-up call for developers and website owners to reassess their protection measures against increasingly sophisticated bots and automated systems.