Cyber Defense Benchmark

The Cyber Defense Benchmark is a dataset designed to evaluate the performance of large language model agents in threat hunting by requiring them to identify timestamps of malicious events from a database of raw Windows event logs, based on real attack procedures and structured within a reinforcement-learning environmen

Papers using Cyber Defense Benchmark (1)

Cyber Defense Benchmark: Agentic Threat Hunting Evaluation for LLMs in SecOps2026