ANALYSIS OF PASSWORD RESISTANCE TO AI-GENERATED ATTACKS

Abstract

Passwords remain one of the key authentication methods; however, their security largely depends on storage methods and the ability to counter modern attacks. Traditional password-cracking techniques, such as brute force and dictionary attacks, are gradually losing effectiveness against more advanced technologies that leverage artificial intelligence.This study analyzes the impact of AI-generated attacks on password resilience and evaluates their effectiveness compared to classical methods.Modern approaches to password generation in real authentication systems and password-cracking mechanisms are examined. Special attention is given to the PassGAN model, which utilizes generative adversarial networks to create passwords with a high probability of matching real user combinations. The effectiveness of such attacks and their ability to bypass standard password complexity verification mechanisms are assessed. An experimental study is conducted, comparing the success rates of AI-generated attacks with traditional methods, such as brute force and dictionary attacks.The experimental results indicate a significant increase in password-cracking efficiency for medium-complexity passwords when using AI-generated attacks. Meanwhile, complex passwords demonstrate significantly higher resistance to such attacks. Based on the obtained data, recommendations are formulated to enhance password security, including the use of longer combinations and the implementation of digital immunity systems capable of detecting and preventing AI-driven threats.The study highlights the importance of an adaptive approach to authentication security and the need for comprehensive solutions that combine strong passwords, multi-factor authentication, and artificial intelligence technologies to protect digital systems. Further development of AI-based attack methods requires continuous improvement of detection and mitigation techniques.