Capture the Bot: Using Adversarial Examples to Improve CAPTCHA Robustness to Bot Attacks

To date, CAPTCHAs have served as the first line of defense to prevent unauthorized access by (malicious) bots web-based services, while at same time maintaining a trouble-free experience for human visitors. However, recent work in literature has shown that sophisticated using advancements Machine Learning (ML) can easily bypass existing CAPTCHA-based defenses. This introduces CAPTURE, novel CAPTCHA scheme based on adversarial examples. Typically examples are used lead an ML model astray. With we attempt make “good use” such mechanisms order increase robustness and security schemes. Our empirical evaluations show CAPTURE produce challenges easy humans solve, time, effectively thwart ML-based bot solvers.