Modern security tools such as firewalls and intrusion detection systems struggle to detect attackers using stolen but legitimate credentials because their actions closely resemble normal user behaviour. This paper proposes an AIpowered cyber deception system that generates and manages dynamic honeytokens, fake credentials, and digital assets with no legitimate use to expose unauthorised access. The proposed architecture integrates AI-based honeytoken generation, a tokenmatching detection engine, real-time alerting, and a monitoring dashboard into one pipeline. As a conceptual, design-stage framework, it aims to enable early, near-zero-false-positive detection of credential misuse and lateral movement.

Anitha V, Makheshkumar R P, Thirunithi T, Saranraj G, Anand S,"AI-Powered Cyber Deception System Using Dynamic Honeytokens for Credential Misuse Detection" International Journal of Advanced Multidisciplinary Research and Educational Development, V2(5): Page(596-600) September - October 2026. ISSN: 3107-6513. www.ijamred.com. Published by Scientific and Academic Research Publishing.