AI Agent Bypassing Human-in-the-Loop Approval Guardrails
Detects AI agents attempting to disable or bypass security guardrails, approval gates, or content filters. Additionally, it identifies high-risk or destructive actions taken by an AI agent that lack the required preceding human-approval audit record.
YARA-L
