The Enterprise AI
Threat Model

A shareable reference for securing the infrastructure connecting your users, models, tools, and data — the trust boundaries, the threat tables, and the control mapping, in one document.

• Trust-boundary model: users → models → tools → data
• Human-workflow threats (unauthorized access, leakage, injection, audit tampering)
• Autonomous-agent threats (delegation, tool execution, runaway actions)
• Control mapping and a defense-in-depth reference architecture