Key Takeaways

1. Agents Need Their Own Proof: As AI agents take risky actions inside business systems, AIUC-1 gives enterprises a way to demonstrate that a specific agent is secure, safe, and reliable.

2. There Are Four Stages to Certification: AIUC-1 certification moves through scoping, technical evaluations, a third-party audit, and certification. The process typically takes several weeks for a defined product or agent.

3. AIUC-1 Certification Is Ongoing: Certified organizations face technical retesting at least every quarter and a full annual re-audit to keep their AIUC-1 certificate current.

Today’s headlines are full of instances of AI agents escaping their boundaries and ingesting proprietary data, exposing it to unauthorized parties. Improper deployment and use of AI has compromised government data, machine learning platform user credentials and data, intellectual property, and other sensitive information. Such instances of inadvertent compromise have exposed the risks of using artificial intelligence and AI-powered tools without establishing clear and effective boundaries.

The lessons learned from these incidents are clear: an AI agent with broad access, too much trust, and too few guardrails can cause a security incident on its own. Organizations are now seeking proven frameworks for adopting AI without putting data security at risk.

Check Your Compliance Status Now!

Don’t know where to start? Answer a few questions and get free, personalized framework recommendations in 1 minute.

CHECK COMPLIANCE REQUIREMENTS HERE

Why AIUC-1 Is Emerging as the Standard for AI Agent Assurance

Agentic AI is different from earlier generations of AI tools. Instead of simply answering questions, agents call tools, query databases, write and execute code, and act autonomously on behalf of users. That shift expands the attack surface to include prompt injections, unauthorized tool use, data leakage, and hallucinated actions with real-world consequences.

Established frameworks such as SOC 2 and ISO/IEC 27001 remain essential, but they were built to evaluate organizational controls, not to test how an autonomous agent behaves under adversarial pressure. Enterprise buyers increasingly want evidence that the agent itself has been tested and can be trusted to operate in their IT estate.

The Artificial Intelligence Underwriting Company (AIUC) developed the Artificial Intelligence Unified Code-1 (AIUC-1) to fill that gap. First released in July 2025, AIUC-1 pairs an independent audit of controls with hands-on adversarial testing of the agent. Adoption has moved quickly: Cursor, ElevenLabs, UiPath, Sierra, and Fin (formerly Intercom) are among the companies that have certified AI products under AUIC-1. AIUC also offers insurance for AI agents, which gives the standard a direct connection to how AI risk is priced and transferred.

What Is AIUC-1?

AIUC-1 is a compliance and assurance framework designed specifically for AI agents. It is often compared to SOC 2 because it gives vendors a recognized, third-party-validated way to answer customer due diligence questions.

However, AIUC-1 is not a regulatory mandate. Instead, it is a voluntary standard that complements broader frameworks such as the NIST AI Risk Management Framework, ISO/IEC 42001, and the EU AI Act.

A defining feature of AIUC-1 is that the standard itself is refreshed quarterly. That cadence is intended to keep requirements aligned with a threat landscape that changes far faster than traditional compliance cycles.

What AIUC-1 Certification Covers

AIUC-1 requirements are organized into six domains:

  • Data and Privacy: Controls that protect sensitive data, limit what agents can access, and prevent data from leaking across users or contexts.
  • Security: Defenses against prompt injection, jailbreaks, adversarial inputs, and unauthorized access to tools or systems.
  • Safety: Safeguards that prevent harmful, inappropriate, or out-of-policy outputs and actions.
  • Reliability: Measures that reduce hallucinations and ensure agents behave predictably and call tools correctly.
  • Accountability: Governance, logging, incident response, and clear ownership of agent behavior.
  • Society: Protections against cyber misuse and broader catastrophic risks.

Importantly, AIUC-1 certification is typically scoped to specific AI products or agents rather than an entire organization. The certificate’s scope statement identifies exactly what was certified, which is something buyers should review closely.

Compliance questions? Get answers!

Book a free 30-minute consultation with a specialist to find your path to compliance. Secure your spot today.

SPEAK TO AN EXPERT

The AIUC-1 Certification Process, Step by Step

AIUC describes the certification process as four major stages: scoping, evaluations, audit, and certification. Depending on an organization’s AI governance maturity, AIUC’s materials suggest first-time certification generally takes about four to 10 weeks.

Stage 1: Scoping: Scoping defines what will be certified. The organization, AIUC, and an accredited auditor agree on which agents, use cases, data flows, and integrations are in scope, and the auditor documents applicable requirements in a Statement of Applicability. The version of the standard used for the engagement is also locked in at this stage.

Scoping is also where gaps surface. Organizations compare existing policies, technical guardrails, and documentation against AIUC-1 requirements and build a remediation plan. Companies that complete a readiness assessment before this stage tend to move through the remaining phases faster.

Stage 2: Technical Evaluations: Next, AIUC tests the agent directly. Its technical evaluations simulate real-world adversarial behavior, typically through 1,000 to 5,000 test scenarios covering risks such as prompt injection, data leakage, harmful outputs, hallucinations, and unsafe tool use. To pass, no critical or major findings can remain open, so organizations should expect a cycle of testing, remediation, and retesting.

Stage 3: Third-Party Audit: While AIUC tests the agent, an accredited audit firm reviews the controls around it. The auditor collects and evaluates evidence of policy, operational, and technical controls across all six domains, then recommends whether certification should be granted.

Stage 4: Certification: Finally, AIUC reviews the auditor’s report and evaluation results and issues the AIUC-1 certificate. Certified organizations receive a certificate valid for 12 months; an audit report with third-party attestation that can be shared with customers on request; and an AIUC-1 badge for trust centers, websites, and sales materials.

An AIUC-1 certification gives enterprises more confidence in the safety of AI agents deployed in their networks.

Ongoing Requirements: Quarterly Retesting and Annual Recertification

Because the AI threat landscape is highly dynamic, AIUC-1 certification evolves to keep pace with change. For that reason, organizations should plan for these continuing obligations:

  • Quarterly Technical Retesting: AIUC repeats technical evaluations at least every three months, so the certification continues to reflect how the agent behaves as models, prompts, and threats evolve.
  • Annual Recertification: A full re-audit renews the certificate each year, beginning roughly two months before expiration, although AIUC describes it as lighter than the initial audit.
  • A Moving Standard: Because AIUC-1 is updated quarterly, controls that satisfied one version may need adjustment for the next, so readiness work should always be checked against the current release.
  • Continuous Evidence: Logging, incident response records, and change documentation should be maintained year-round rather than assembled just before the audit.

Navigating AIUC-1 With Experienced Guidance

AI technology, AI risk, and AIUC-1 are all evolving at a pace few compliance programs have faced before. New agent capabilities introduce new attack paths, quarterly updates to the standard shift the target, and the cost of a misstep can be significant.

Working with an outside firm that has deep experience in AI and security assessments can help organizations move through this environment with confidence.

An experienced partner can help define a realistic certification scope, perform a candid gap assessment, strengthen governance and technical controls, prepare evidence, and keep controls aligned with each new release of the standard. Because audit impartiality rules generally prevent the certifying auditor from also preparing an organization for its audit, independent readiness support is especially valuable.

IS Partners brings years of experience in security assessments, compliance, and AI governance to help organizations prepare for emerging standards like AIUC-1 and build AI programs that customers can trust.

What Should You Do Next?

  1. Inventory Your AI Agents: Identify every AI agent in production or development, along with the data, tools, and systems each one can access.

  2. Assess Your Readiness: Compare your current AI governance, security controls, and documentation against AIUC-1’s six domains to uncover gaps before engaging AIUC and an accredited auditor.

  3. Engage an Experienced Partner: Contact an outside firm with AI and security assessment expertise to build a practical roadmap toward AIUC-1 certification and ongoing compliance.

About The Author

Get started

Get a quote today!

Fill out the form to schedule a free, 30-minute consultation with a senior-level compliance expert today!

ioc-checkAnalysis of your compliance needs
ioc-checkTimeline, cost, and pricing breakdown
ioc-checkA strategy to keep pace with evolving regulations

Great companies think alike.

Join hundreds of other companies that trust IS Partners for their compliance, attestation and security needs.

Specialty_Capital_Logonlex-logoDHEC_report_logomcl logoXL_net_623x538_transparent_Website_Featureteladoc

Scroll to Top