Guardrail Actors

What are Guardrail Actors?

Guardrail actors are individuals, groups, or entities responsible for carrying out malicious activities targeting your AI agents, MCP endpoints, and autonomous systems. They are often motivated by financial gain, espionage, ideology, or disruption. Guardrail actors targeting agentic systems may attempt prompt injections, tool abuse, data exfiltration through agent interactions, or unauthorized access to agent capabilities. They can range from lone hackers to organized crime groups, nation-states, or insider threats.

Key Metrics to Monitor

Critical Actors: Guardrail actors posing a high risk to your agentic systems due to their capability and intent

Active Actors: The total number of guardrail actors currently engaging in malicious activities against agent components

Guardrail Activity Timeline: Tracks the frequency and volume of attacks over time (e.g., agent component invocations, tool abuse attempts)

Guardrail Actor Map: Geographically maps the origin of guardrail actors targeting your autonomous systems