AIUC-1 Certification

The standard for AI agent security, safety and reliability.

AIUC-1 unlocks enterprise adoption. Covers all risks. Grounded in technical testing. Built with 250+ Fortune 500 security leaders.

Begin Certification View Consortium

Consortium Members

Phil Venables
Partner, Ballistic Ventures; fmr. CISO, Google Cloud, & fmr. CISO, Goldman Sachs

Case Studies

Certified Agents

AIUC-1 covers all agents. For companies that want to stand out on security.

Cursor Certified with AIUC-1

Why They Certified

“AIUC-1 is the only standard that keeps up with our pace.”

Kenneth Thomas Moras, Security, Cursor

View Full Case Study

Domains

All Enterprise Risks

AIUC-1 covers all risks that slow enterprise adoption. Each domain includes technical and operational controls. Controls are tested with red-teaming and independent audits.

- Data Leaks
- PII Protection
- IP Infringement
View Requirements
- Prompt Injection
- Jailbreaks
- Unauthorized Actions
View Requirements
- Harmful Outputs
- Red-Teaming
- Human Oversight
View Requirements
- Hallucinations
- Tool Calls
- Agent Reliability
View Requirements
- Incident Response
- Logging
- Governance
View Requirements
- Cyber Misuse
- Catastrophic Risk
View Requirements

Use Cases

How Companies use AIUC-1

Certify the AI agents built by your organization.

Begin Certification

And more

Customer Support Agent

Red-Teaming

Evals

2402 / 3215

P3 Failure

Technical Controls

D001 Prevent Hallucinated OutputsB001 Require Exception Escalation

Policy Controls

D001 Prevent Hallucinated OutputsB001 Require Exception Escalation

Legal Controls

D001 Prevent Hallucinated OutputsB001 Require Exception Escalation

For Buyers

Evaluate AI Products

Ask your AI vendors if they are AIUC-1 certified. The 50+ page independent audit report covers how guardrails are implemented and red-teaming results.

For Security Teams

Build to AIUC-1 spec

Merge AIUC-1 into your control library. Let your product teams build to a clear security spec. Crosswalked to all major AI frameworks.

Explore Controls

Certification Process

No Guesswork

Four steps: scoping, evals, audit, certification. Scoping starts with a gap assessment, ends with crisp work plan. You always know what's next and who owns it.

Begin Certification

Scoping

Define which agent or system is being certified, the deployment context, and the relevant control domains. Gap assessment in 48 hours. Organizations are prepped to pass Red-Teaming against real-world risks and implement relevant AIUC-1 Controls.

AIUC-1 Creates Custom Tests
Locks in your AIUC-1 Standard Version for 1 Year
Agent Configuration and Interface Bounds

Evals

AIUC conducts thousands of real‑world scenarios testing against jailbreaks, prompt injection, data leakage, hallucinations, and unsafe tool calls. Red-teaming typically involves 1,000 to 5,000 different test scenarios.

Organization provides representative agent configs
AIUC-1 creates custom tests and conducts red-teaming
Organization Implements Controls
Quarterly retests on your AIUC-1 Standard Version

Consortium

Built with Security Leaders

The consortium unites 250+ Fortune 500 security leaders. They protect trillions in payments, sensitive medical data, and military systems. Together, we set the standard.