AIUC describes AIUC-1 as a standard for AI agents covering data and privacy, security, safety, reliability, accountability, and societal risk. Witnora does not certify AIUC-1 compliance. It can produce technical evidence that helps a team prepare for AIUC-1-style review.
| AIUC-1 concern | Witnora evidence |
|---|---|
| Data and privacy | MCPBench taint tracking, canary leakage checks, Tripwire sensitive-text assertions. |
| Security | MCP/tool poisoning scenarios, prompt-injection browser tests, runtime policy gates. |
| Safety | Runtime approval requirements for high-risk SUBMIT, PAY, SEND, and UPDATE actions. |
| Reliability | Deterministic CI gates, repeatable scenarios, pass/fail reports, trace artifacts. |
| Accountability | Onegent Runtime audit packets with action, approval, execution, verification, and event timeline. |
| Reviewability | agentcert-evidence.json and agentcert-report.md as portable evidence artifacts. |
Witnora should keep mapping evidence to AIUC-1 concerns while avoiding claims of official certification. The target phrase is:
Witnora helps generate reviewable evidence for AIUC-1-style agent assurance.