AI COMPLIANCE COMPARE

Dispatch · standards, AI governance · 2026-11-04

SOC 2 plus AIUC-1: the AICPA's joint AI assurance offering, and which platforms describe anything that maps to it

By the Signal Desk, AI Compliance Compare · Published · 5 min read

Short answer

On 19 October 2026 the AICPA and the Artificial Intelligence Underwriting Company (AIUC) announced a joint assurance offering: one CPA-led engagement that produces a SOC 2 report and an AIUC-1 certification, described by the AICPA as the first comprehensive independent review of business risk related to AI systems and agents. Several CPA firms are delivering it. For a company that ships AI agents, this is a new thing a customer may ask for, and a new line to check on a compliance platform's framework page.

What was announced?

A joint certification that combines a System and Organization Controls (SOC 2) report with an AIUC-1 certification, the standard AIUC describes as covering AI agent safety, security and reliability. The AICPA's release stresses that CPAs lead the engagement, which keeps the SOC 2 part under the AICPA attestation standards buyers already recognize. The Journal of Accountancy summarizes it as "one CPA-led engagement that provides assurance on AI". The release says several CPA firms are now delivering the joint offering to clients.

Who would ask for it?

The same buyers who ask for SOC 2 today, once the product they are buying includes an agent that takes actions. A procurement team that accepts a SOC 2 report for data handling has had no equivalent document for "what can your agent do, and what stops it doing the wrong thing". The joint offering is an attempt to give them one without inventing a new audit profession. Expect the request to arrive first in regulated industries and in enterprise deals where an AI feature touches customer data.

How does it land on a compliance platform?

Three ways. First, as a framework. Vanta's framework page lists ISO 42001, NIS2 and DORA, and Sprinto's lists 200+ digitized frameworks; neither listed AIUC-1 on the pages we read in September, and we have not re-read them for this post. Ask whether AIUC-1 is mapped and whether the mapping is maintained by the vendor. Second, as scope. Your platform's agents are AI systems inside your environment, so a joint engagement would reasonably cover how you govern them, which returns to the questions in our 3 November post on the proposed 2027 criteria. Third, as evidence. An auditor doing the joint engagement will want logs of agent actions; Scytale's MCP release describes per-call logging, and that is the kind of artifact to ask every vendor for.

What does this mean for AI governance frameworks generally?

It is the second signal in a fortnight that SOC 2 is being pulled toward AI. The exposure draft of 28 October adds AI points of focus to the criteria themselves; the AIUC-1 pairing adds a certification alongside the report. ISO 42001 already exists for AI management systems. A company that sells AI should expect to hold at least two of the three within a couple of years, and should pick a platform whose cross-mapping lets one set of controls serve all of them. Our framework ladder page shows which vendors describe control cross-mapping on their own pages.

What did we change on this site?

Nothing in the scores. We have added AIUC-1 to the field guide on AI governance frameworks with a short description and the AICPA source, and added "Is AIUC-1 mapped?" to the demo questions on the framework ladder page.

Questions buyers ask

What is AIUC-1?

A standard from the Artificial Intelligence Underwriting Company for AI agent safety, security and reliability. Since 19 October 2026 the AICPA and AIUC offer a joint CPA-led engagement that pairs an AIUC-1 certification with a SOC 2 report.

Do compliance platforms support AIUC-1?

None of the vendor framework pages we read on 29 September 2026 listed it. Ask the vendor whether it is mapped and maintained.

Sources

Keep reading

More from Dispatch