Safety visual guide

Tool and data trust boundaries

Where identity, policy, approval, validation, and audit controls belong around an AI tool call.

Editable sourceMermaidOpen full-size SVG ↗Download .mmd source

A trust-boundary diagram shows a User and Human approval connected to an AI host containing the Model and policy context. The host calls a Tool gateway that performs schema validation, authorization, and result validation before reaching Downstream systems. Identity service and Audit log connect to the host and gateway. — Safe tool use layers host policy, typed validation, authorization, human approval, downstream enforcement, and audit evidence.

Read the visual

What this shows

A model is one participant in a larger system. The host supplies identity and policy, a tool gateway validates typed input and checks authorization, a human approves higher-impact actions, downstream systems enforce their own permissions, and audit evidence captures what actually happened.

Carry this forward

Key takeaways

Never treat the model as the authorization boundary.
Validate inputs before execution and results before returning them to the model.
Use human approval for consequential actions and preserve an auditable record.