Insider Threat Matrix™Insider Threat Matrix™
HumanSynthetic
Detections
Preventions
Github
Synthetic

In partnership with Above Theory

Directive

The objectives and constraints that determine how a synthetic subject behaves.

Configuration

The access and settings that define what a synthetic subject can do.

Autonomous Action Control Enterprise Retrieval Access Model and Build Provenance Model Objective Alignment Persistent Memory Access Standing Instruction Stack

Invocation

The inputs or triggers that cause a synthetic subject to act.

Agent-to-Agent and Tool-Output Invocation Deconstructed and Staged Invocation Indirect Untrusted-Content Invocation Memory-Resident Invocation Triggered and Delayed Invocation

Adverse Outcome

The outcome that harms or undermines an organization.

Agentic Harm Propagation AI-Mediated Fraud and Misappropriation Destructive System or Data Action Erroneous Autonomous Action Harmful or Non-Compliant Output Identity Misattribution and Impersonation Harm Sandbox Escape and Out-of-Boundary System Access Unauthorized Log or Record Change Unbounded Resource Consumption

Opacity

The conditions that frustrate observation, explanation, attribution, or containment of a synthetic subject.

Deliberately Concealed or Misleading Reasoning Evaluation-Aware Behavior False Operational Self-Reporting Reproducibility and Containment Gaps Source Provenance Obfuscation Structurally Unreliable Reasoning Trusted Channel Camouflage

213 knowledge objects. 213 recently added.

In partnership with Above Theory