AuditBench is a new benchmark of audit logs from 50+ malicious and benign scenarios that evaluates five LLMs on four security investigation tasks and analyzes their performance and error profiles.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
fields
cs.CR 2years
2026 2verdicts
UNVERDICTED 2representative citing papers
Sieve uses an LLM to generate executable queries from natural language security questions grounded by auto-extracted log-format context, cutting error rates over 3x on complex temporal and cross-event tasks versus manual scripting across 133 queries and 5 log types.
citing papers explorer
-
Benchmarking and Exploring the Capabilities of LLMs for Attack Investigations
AuditBench is a new benchmark of audit logs from 50+ malicious and benign scenarios that evaluates five LLMs on four security investigation tasks and analyzes their performance and error profiles.
-
Parser-Free Querying of Security Logs
Sieve uses an LLM to generate executable queries from natural language security questions grounded by auto-extracted log-format context, cutting error rates over 3x on complex temporal and cross-event tasks versus manual scripting across 133 queries and 5 log types.