Agentic AI Shatters Decades-Old Security Assumptions

A new paper from arXiv introduces the 'Infinite Impostor' attack model, where autonomous AI agents hijack trusted relationships by producing high-fidelity, individually tailored deception at mass-market scale. The authors argue that this collapses the traditional tradeoff between deception fidelity and deployment scale, exhausting a security paradigm rather than merely intensifying threats. The research was published on May 26, 2025, under arXiv ID 2605.16436.