SOC 2 + AIUC-1: Comprehensive Assurance for Agentic Software

Cyber and AI risk assurance in a single examination to close the gap between traditional cyber assurance and the realities of autonomous AI. Developed in collaboration with the AICPA and AIUC and delivered by Schellman.

If your company is deploying externally facing or high-impact AI agents that have access to data, tools, systems, or decision-making authority, proving trustworthiness to customers, boards, and regulators has become both increasingly complicated and important.

SOC 2 remains an essential pillar of trust, but as enterprise buyers increasingly ask how AI agents actually behave under pressure, those questions call for a complementary layer of assurance built specifically for agentic systems.

That’s what a SOC 2 + AIUC-1 report is designed to deliver.

What is SOC 2 + AIUC-1?

A SOC 2 + AIUC-1 report is cyber and AI risk assurance delivered in a single examination, developed in collaboration with the AICPA and AIUC, and delivered by Schellman. It produces one opinion, signed by a licensed CPA firm, covering both criteria sets: the AICPA Trust Services Criteria (SOC 2) and the agent-focused AIUC-1 framework.

SOC 2 + AIUC-1 is comprehensive assurance for agentic software as a way to prove the systems around your AI and the AI's own behavior have been independently tested.

Who is a SOC 2 + AIUC-1 Report For?

SOC 2 + AIUC-1 is built for companies deploying externally-facing or high-impact AI agents with access to data, tools, systems, or decision-making authority. This includes coding agents, chatbots, automation agents, and systems where multiple agents work together.

SOC 2 + AIUC-1 is most relevant for organizations with AI products that are:

• Agentic: the workflow involves at least one LLM-powered step
• Clearly scoped: built around a defined set of capabilities within the testing boundary
• Exposed to real risk: typically in production, commercially facing, and/or operating in a regulated space

AI category leaders already certifying include Cursor, Lovable, Harvey, UiPath, Collibra, Fieldguide, and Sierra.

Why is a SOC 2 + AIUC-1 Report Important?

SOC 2 alone wasn't designed to evaluate autonomous AI behavior. The AICPA Trust Services Criteria covers infrastructure, access controls, and change management for the systems around your AI, but they don't address agent behavior. When an LLM has credentials, can call tools, and take autonomous actions as a decision-maker, the risks are different.

A SOC 2 report validates the infrastructure, access controls, and operational rigor around your systems, and it remains essential on its own for organizations that need that assurance. AIUC-1 takes that assurance a step further, with adversarial testing and stress testing designed specifically to evaluate how an AI agent acts once it has real autonomy across credentials, tool access, and decision-making authority.

That additional layer is why a SOC 2 + AIUC-1 report has become more important in the past year than ever before. AI risk doesn't move at the pace of a traditional annual audit. Models, capabilities, attack techniques, and real-world incidents all shift faster than a once-a-year review can track. AIUC-1 controls are retested quarterly, keeping certification aligned to the current threat landscape rather than a stale, point-in-time snapshot.

Just as importantly, the certified boundary is stated in the report, not inferred from a badge. Every SOC 2 + AIUC-1 report specifies exactly what was tested, with no ambiguity for a prospect's security or legal team to guess at or interpret.

How SOC 2 and AIUC-1 Work Together

SOC 2 provides infrastructure and operational assurance. AIUC-1 technically tests how AI agents actually behave through adversarial testing and quarterly retesting against a defined set of AI capabilities in scope. Together, they close the gap for comprehensive compliance: infrastructure assurance, AI governance, and behavioral validation in one engagement.

Combined, the report expands risk coverage specifically for agentic AI systems, across four key areas:

• Data: unauthorized access, customer data confidentiality, personal data and IP leakage, cross-customer exposure
• System Integrity: change management failure, processing integrity, insecure generated code, availability
• Agent Behavior: prompt injection, unauthorized tool calls, hallucinated outputs, out-of-scope outputs
• Accountability: logging and traceability, human review of high-risk actions, model supply chain, regulatory documentation

What Does a SOC 2 + AIUC-1 Report Give You?

SOC 2 + AIUC-1 gives you one single consolidated report deliverable instead of having to juggle disconnected conversations around security and AI governance. This one report answers both the infrastructure and agent-behavior questions upfront, resulting in fewer security questionnaires and follow-up diligence calls.

You also receive a defined, stated assurance boundary with no guesswork about what was covered. The report involves independently tested agent behavior, resulting in a competitive edge in enterprise sales cycles and a faster path through enterprise and regulated industry reviews.

What is the SOC 2 + AIUC-1 Report Timeline?

The streamlined process is typically completed in as little as two months, from start to finish, running alongside your SOC 2 engagement rather than adding a separate three-month track. Indicative customer staffing requirements are roughly 20% of one GRC resource over the engagement period.

The SOC 2 + AIUC-1 timeline typically involves:

1. Scoping (1-2 weeks): products and criteria confirmed
2. Audit Readiness (2-4 weeks): evidence and test environment checked
3. Technical Evaluations (3-5 weeks): adversarial testing and retest
4. Evidence Review (1-2 weeks): controls tested
5. Report and Certificate (1-2 weeks): SOC 2 + AIUC-1 attestation report issued by Schellman, AIUC-1 certificate issued by AIUC

Getting started is low commitment, with each step producing a clear decision before any engagement letter. The process to get started typically follows:

1. Consultation (30 minutes, Schellman and AIUC): confirm agents in scope, current posture, indicative timeline
2. Gap assessment (outside-in, no evidence requests): confirm requirements met or likely met, and what technical setup is needed
3. Timeline review: proposed engagement timeline, fieldwork booked, sequenced to internal milestones

Why Schellman for a SOC 2 + AIUC-1 Report

Schellman has spent over two decades at the intersection of cybersecurity, data privacy, and emerging technology risk auditing the infrastructure, controls, and now the AI systems that the world's most scrutinized companies run on. We work with leading LLM providers, hyperscalers, and category-defining AI companies, giving us a front-row view into how the most sophisticated AI organizations build governance, security, and trust into their products.

We were the first ANAB-accredited ISO 42001 certification body, and the first authorized assessor for AIUC-1, because our mission is to evolve alongside and ahead of emerging risk, not wait for a standard to mature before helping clients prove they're trustworthy.

Independence matters here more than ever. Only independent CPA firms can conduct SOC 2 + AIUC-1 assessments. When boards, customers, and regulators read a Schellman report on your AI agents' behavior, they can trust it reflects what you actually built.

If you're ready to find out where you stand or learn more about pursuing a SOC 2 + AIUC-1 report, contact us today to start the conversation.