ThreatHunter.ai Releases MILBERT 2.1 to Enhance Identity Attack Detection
ThreatHunter.ai has launched version 2.1 of its identity detection engine, MILBERT, designed for hybrid Active Directory and Entra ID environments. The update introduces continuous user-behavior tracking that identifies attacks occurring after a successful login—such as adversary-in-the-middle phishing or session theft. Unlike traditional systems, it does not assume the legitimacy of multi-factor authentication (MFA) completion but evaluates sessions and sequences to detect anomalies. The new version uses two models: one scores each authentication event using gradient boosting while another fine-tuned language model analyzes a window of events for potential attack chains. MILBERT 2.1 can label threats, provide rationales, and recommend actions like Allow or Block without altering directory state directly. Training data included real-world MFA theft cases and user behavior observations. MILBERT 2.1 is capable of processing over 74,000 authentication events per second with high accuracy on internal infrastructure—important for clients under CMMC and FERPA regulations who cannot send identity telemetry to public cloud services.
Tickers: $SYK
Topics: Active Directory Entra ID identity security MILBERT 2.1 phishing detection