Trust Tier
Graduated levels of trust assigned to AI agents: low trust requires explicit consent for every tool, high trust auto-approves common patterns. Capped by the user's RBAC permissions.
A trust tier is a graduated level of automation granted to AI agents, ranging from "requires explicit consent for every tool call" to "auto-approves common patterns within RBAC boundaries." Trust tiers are set per-user and capped by the user's RBAC permissions — a user with viewer access cannot grant an agent write-level trust.
Common trust tier levels
- Low / Sandbox: every tool call requires explicit user consent
- Medium / Standard: auto-approve non-destructive operations; require consent for writes
- High / Trusted: auto-approve all operations within the user's RBAC permission set
Access-aware capping
The trust tier cannot exceed what the user's RBAC actually allows. If the user cannot delete invoices, no trust tier setting can grant the agent that ability. Trust tiers are a convenience layer on top of mandatory RBAC enforcement — not a bypass.
Related terms
Security pattern requiring a human reviewer to approve sensitive operations before an agent executes them. Single-use approval consumed on the next matching call.
User-facing interface where a human explicitly authorizes an AI agent to access specific tools within defined boundaries. Time-boxed, revocable, and access-aware.
Technique that identifies and monitors AI agents by behavioral signature. Detects identity changes, tool usage pattern shifts, and privilege drift before they become incidents.