According to a leak by X platform blogger Lain on the Blockchain on June 15, Google is testing a new set of reCAPTCHA human verification mechanisms: users must grant camera permissions to the webpage, record a short video of waving to the lens, and the system then uses computer vision to extract coordinate points from 21 key points such as finger joints and palm positions to determine whether the user has completed the specified gesture. Google says this move is aimed at combating cyber fraud such as batch registration of bots and credential stuffing attacks. The recorded video is not linked to the user’s identity and does not record audio; it is automatically deleted after verification.
This solution has drawn criticism from some users, who argue that it further pushes the Internet toward biometric authentication, and even though the data is deleted immediately and not tied to identity, users still need to enable their cameras to complete verification when visiting a website, which involves significantly higher friction costs compared to existing text recognition, image selection, and invisible reCAPTCHA. The system is still in the testing phase, and Google has not yet announced an official launch timeline.