Show HN: Movement-based captcha – temporal reasoning to foil AI vision (github.com)

A new proof-of-concept captcha has been introduced that employs movement as a means to differentiate humans from AI vision tools, addressing the growing trend of AI solutions easily solving traditional static captchas. This innovative approach utilizes temporal reasoning, requiring users to identify characters that move along specific geometric paths while decoy characters follow unrelated patterns. This method makes it significantly more challenging for AI to analyze because motion combined with spatial awareness complicates frame-by-frame interpretation. The significance of this movement-based captcha lies in its potential to enhance security against AI-generated submissions by pushing the boundaries of what current vision models, including advanced ones like GPT-4V and Claude, can handle. The prototype functions entirely on the client-side and highlights that a static frame offers no useful information, emphasizing the need for dynamic interactions. However, concerns regarding accessibility exist, as these challenges may be difficult for individuals with vestibular disorders or for those relying on screen readers. The idea shows promise but requires further evaluation for security and usability before broader implementation.