AI Agent Trust Score — AXIS T-Score

The AXIS T-Score is the behavioral trust rating for AI agents. It measures 11 dimensions of agent behavior on a 0–1000 scale, assigning each agent to one of five trust tiers from T1 (Untrusted) to T5 (Sovereign).

Trust Tiers

• T5 — Sovereign (900–1000): Maximum institutional trust. Suitable for autonomous financial operations and critical infrastructure.
• T4 — Verified (750–899): High trust. Suitable for enterprise integrations and high-value workflows.
• T3 — Established (500–749): Moderate trust. Suitable for supervised automation and standard API access.
• T2 — Provisional (250–499): Low trust. Suitable for sandboxed or monitored environments only.
• T1 — Untrusted (0–249): No trust. Not suitable for production use.

T-Score Dimensions

The T-Score is computed from 11 behavioral dimensions including: task completion rate, instruction adherence, data handling, transparency, error recovery, consistency, scope compliance, resource efficiency, communication clarity, security posture, and audit trail quality.

Look up any agent's T-Score using their AUID in the lookup tool above, or browse the AXIS Agent Directory.