Published on November 11, 2026 2 min read

Compliance Briefs: AIUC-1 Certification for AI Agent Testing Can Now Pair with SOC 2

A senior tech lead or manager mentoring a diverse team of junior developers in a modern office. Concept for project management, corporate training, and professional growth.

An overview

The American Institute of CPAs (AICPA) and the Artificial Intelligence Underwriting Company (AIUC) announced on October 18, 2026 a new joint offering that gives organizations an independent review of business risk tied to AI systems and agents.

At the center of the offering is AIUC-1, a certification standard for the security, safety, and reliability of AI agents. Under the new arrangement, an AIUC-1 certification can be paired with a SOC 2 report and delivered through a single CPA-led engagement.

What is an AIUC-1 certification?

For companies that build or use AI-powered tools, an AIUC-1 certification offers external validation that the tools touching their data are working as intended and provides a clearer view of AI risk.

Unlike a policy review, AIUC-1 doesn’t stop at how AI agents should behave. It tests how they actually behave in practice. Its controls cover six principles and maps them to established compliance frameworks:

  • Data and privacy
  • Security
  • Safety
  • Reliability
  • Accountability
  • Society

What happens during an AICU-1 testing?

Agents are tested across hundreds of combinations of risks and attacks, including:

  • Hallucinations
  • Data leakage
  • Manipulative requests
  • Prompt injections

A typical evaluation covers thousands of scenarios. The most serious findings must be fixed before a certificate can be issued. Certification is renewed annually, with technical retesting every quarter.

Is an AIUC-1 certification right for your business?

As new certifications emerge and evolve, they won’t immediately impact every company. If you answer yes to any of the questions below, AIUC-1 may be worth a closer look.

  • Do you sell AI products to larger companies or highly regulated/high-risk industries, such as healthcare or financial services?
  • Are AI questionnaires slowing down your deals?
  • Are you using AI in high-visibility settings, such as customer service, sales, or claim adjudication, where one mistake could damage trust?
  • Do your AI tools have access to sensitive data or act on their own as agents?

If your AI use is limited to low-risk internal tools, strong existing controls may be enough for now.

What should businesses do now?

A full launch of the joint SOC2 + AIUC-1 certification is expected in early 2027.

Have questions about whether AICU-1 certification belongs on your roadmap? Aprio’s Risk Advisory and Assurance Services (RAAS) team can help you understand what these changes mean for your business.

Connect with us