Cognitive CAPTCHA authorization codes

Abstract

CAPTCHA (Completely Automated Public Turing Test to Tell Computers and Humans Apart) has long been an essential mechanism for protecting online services against automated abuse. By leveraging Human–Artificial Intelligence (HAI) interactions, CAPTCHAs aim to differentiate legitimate users from bots. Over time, CAPTCHA schemes have evolved from simple text- and image-based puzzles to more advanced mechanisms intended to improve usability and resilience. However, conventional CAPTCHAs have been compromised by automated solvers, while even modern approaches remain vulnerable to human-assisted relay attacks.

The advent of fourth-generation bots, capable of closely mimicking human behavior, presents a major challenge to CAPTCHA design. Cognitive-based CAPTCHAs demonstrate strong resistance to automated attacks, but their reliance on limited challenge sets and trusted hardware makes them susceptible to relay-based threats. Despite these drawbacks, cognitive approaches offer a promising foundation for advancing human-centered cybersecurity.

This study investigates the development of next-generation CAPTCHA systems with the following objectives:

Analyze the limitations, challenges, and opportunities of current CAPTCHA schemes.

Strengthen resilience by integrating cognitive features with advanced machine learning methods such as Convolutional Neural Networks (CNNs) and Deep Neural Networks (DNNs).

Establish robust evaluation methods to measure usability and the effectiveness of user–CAPTCHA interaction.

Design novel cognitive CAPTCHA authorization codes that provide secure and user-friendly verification while resisting automated and relay-based attacks.

By addressing these objectives, this research contributes to the development of CAPTCHA schemes that balance usability and security, offering stronger defenses against increasingly sophisticated automated and human-assisted threats.