Adaptive Two-Factor Authentication System and Method Based on AI-driven Risk Assessment with Out-of-Band OTP Transmission

Abstract

The invention provides an adaptive two-factor authentication (2FA) system that dynamically generates one-time passwords (OTPs) based on real-time, AI-driven risk assessment. The system collects contextual metadata, including time, IP address, geographic location, device and browser profiles, and user history, and computes a risk score using a trained machine-learning model. According to the risk score, OTP properties are adjusted, including length, character set and complexity, validity period, and number and type of out-of-band channels. An independent out-of-band (OOB) server, physically and logically separated from the primary authentication server, generates and delivers the OTP via selected out-of-band channels such as a mobile application or email, and may trigger an administrator alert. Under high-risk conditions the system strengthens OTP policies and shortens validity, thereby improving resistance to phishing, man-in-the-middle, replay, and session-hijacking attacks while balancing security with user convenience.

Dependencies are inferred from the claim text.