NONLINEAR ADAPTIVE TRUST MODELING FOR AGENTIC AI-ORCHESTRATED ZERO-TRUST MANAGED FILE TRANSFER SYSTEMS WITH PASSWORDLESS AUTHENTICATION INTEGRATION

Abstract

The widespread move away from password-based authentication toward biometric and device-bound cryptographic identity is forcing a fundamental rethink of how enterprise data-transfer pipelines establish and maintain session trust. Managed File Transfer (MFT) systems sit at the centre of this challenge: they move sensitive, regulated data in bulk, yet most deployed platforms still authenticate users with static credentials that offer no defence against phishing, replay, or slow-drift insider attacks. This paper proposes the Nonlinear Adaptive Trust Modeling (NATM) framework, which couples passwordless biometric authentication with a mathematically rigorous, agentic AI-orchestrated zero-trust architecture for MFT security. The framework models per-entity trust as the solution of a nonlinear stochastic ordinary differential equation (ODE) whose evidence function is fed directly by ScrambleID cryptographic identity tokens each token carrying a biometric hash, device fingerprint, and geolocation risk signal. An agentic reinforcement-learning (RL) policy network continuously integrates the ODE and autonomously issues per-session decisions permit, step-up re-authentication, or deny without any human-in-the-loop delay. A Lyapunov stability argument shows that the trust system converges to a bounded attractor under stochastic perturbation, which provides a formal service-level guarantee that purely data-driven baselines cannot offer. Experiments on three calibrated enterprise workloads show a breach detection rate of 94.7%, a false-positive rate of 1.2%, and a mean trust-update latency of 83 ms substantially better than both rule-based and static ML-classifier benchmarks while reducing unnecessary step-up authentication events by 83% relative to the stateless passwordless baseline of Chellu [6]. The results establish that combining differential-equation trust dynamics with passwordless identity evidence and agentic orchestration produces a qualitatively new level of security assurance for enterprise file-transfer operations.