AICPA and Artificial Intelligence Underwriting Company launched a joint assurance certification program. It combines a SOC 2 report with AIUC-1 certification. The program offers independent review of AI system and agent risks. It targets risks like prompt injection, data leakage, and unauthorized actions. Technical testing uses tailored scenarios for each agent’s intended work. Weaknesses found in testing must be fixed and retested before certification. Quarterly retesting and annual renewal are required. The initiative aims to help tech companies prove AI system reliability to business users. Pilots are underway for a Q1 2027 launch.
What changed
AI assurance focused only on governance and compliance without testing agent behavior.
What it unlocks
Independent verification that AI agents are secure, safe, and operating reliably.
- Q1 2027 program launch target
- quarterly technical retesting
- annual renewal for SOC 2 and AIUC-1
What you need to act on it
- pilot phase completion
Sources