Chandrakala Chandrakala

Role model identity contract enforcement as a structural determinant of identity drift suppression in autonomous large language model agents

Autonomous agents powered by large language models (LLMs) are deployed widely across finance, healthcare, legal review, and customer support. Each deployment assumes that an agent will remain within its assigned role—an assumption that adversarial users routinely challenge through identity drift attacks. This work asks...