Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Explore Google's groundbreaking AI system for detecting malicious insider threats in this 38-minute conference talk from Black Hat USA 2025. Learn about FACADE (Fast and Accurate Contextual Anomaly DEtection), an internal security system that has successfully protected Alphabet by scanning billions of events daily over seven years. Discover how this novel self-supervised machine learning system detects suspicious actions by analyzing contextual information surrounding each user activity, utilizing a custom multi-action-type model trained on corporate logs including document accesses, SQL queries, and HTTP/RPC requests. Understand the innovative contrastive learning strategy that overcomes the challenge of scarce incident data by relying solely on benign data for training. Examine the system's clustering approach that enhances detection robustness and achieves exceptional accuracy with false positive rates below 0.01% overall and as low as 0.0003% for single rogue actions like illegitimate access to sensitive documents. Gain practical insights into implementing insider threat detection in your own organization through demonstrations of the newly released open-source version of FACADE, presented by a team of Google security researchers and engineers.
