Questions
Questions about independent AI audits
Straight answers for AI companies heading into a buyer’s security review. Prices and scope are on the audit page, and the comparison page sets an AI audit beside SOC 2, a pen test, ISO 42001 and AIUC-1.
About the audit
What is an independent AI audit?
An independent AI audit tests the AI inside your product as it actually runs, done by a firm that didn’t build it and doesn’t sell you models. Autogeny tests for failure first: whether your AI ever produces a failure you defined as catastrophic (pass or fail), using a targeted red team, and how often it produces unacceptable output (a failure score). Second, it measures how well your AI does its job (a capability score) and whether a cheaper model would do as well. It takes three to four weeks and ends with a report your buyer’s security reviewer can read.
How is an AI audit different from SOC 2, a pen test or ISO 42001?
SOC 2 checks your company’s security controls, a pen test looks for exploitable flaws in your application and infrastructure, and ISO 42001 certifies how you manage AI. None of them tests whether your AI gives wrong answers, leaks one customer’s data to another, or follows an instruction hidden in a document. An AI audit tests exactly that, in a test copy of your running system. Many AI companies selling to enterprises need both SOC 2 and an AI audit, because they answer different questions. The comparison page sets them side by side.
Is an Autogeny audit an AIUC-1 certification?
No. Autogeny is not an accredited AIUC-1 auditor, and the report doesn’t certify you. The audit includes an AIUC-1 readiness check: it reviews every AIUC-1 requirement, maps our test evidence to the ones it covers and lists the likely gaps, so you know where you stand before you pursue certification. Where a requirement needs evidence our tests don’t produce, such as an external pen test of your AI endpoints (B004) or database tenant isolation (A005), the check lists it as a gap and says who usually supplies it. Our free AIUC-1 readiness checklist lists every requirement and the evidence a reviewer asks for.
What do we get at the end of an audit?
You get a report with the catastrophic-failure results and the failure score, including targeted red-team tests run in a test copy of your system: prompt injection, including instructions hidden in content your AI reads; your AI showing one customer another customer’s data; and agent actions that add up to harm. Then the capability score, for how well your AI does its job, and a model cost comparison that shows whether a cheaper model would do as well. It also includes an AIUC-1 readiness check and a policy gap list. Findings are mapped to the AI questions in the security questionnaire your buyer sends you, the CSA AI Controls Matrix and the OWASP Top 10 for LLM Applications. You also get a short buyer summary to attach to a questionnaire, and the test harness, kept ready for monthly testing. The sample report shows the core sections: verdict, failures, results, scope, method, recommendation and re-audit.
Does the audit include a red team or code review?
It includes a targeted red team: adversarial tests built around your AI’s job and run inside a test copy of your system, covering prompt injection, jailbreak attempts against your guardrails, data leaks and harmful agent actions. A broad automated attack run (large open-source attack libraries, with every hit reproduced before we report it) is an optional add-on, sold only with an audit: $5,000 with a workflow audit, $7,500 with an agent audit. We don’t do code security reviews or pen tests of your app; we can point you to independent firms, and we take no referral fee.
How much does an AI audit cost, and how long does it take?
An Autogeny audit costs $20,000 for one AI workflow (about three weeks) or $27,000 for a tool-using agent (about four weeks), as a fixed fee, paid 50% at signing and 50% on delivery. Monthly testing is $2,500 a month per AI workflow or $3,750 a month per tool-using agent, with a six-month minimum. Paid annually, it is 15% off: $25,500 or $38,250 a year. A re-audit after you move to a new model or make a major change is 40% of the audit fee. For comparison, a first-year SOC 2 Type 2 program typically costs $30,000 to $80,000.
What’s the difference between an AI workflow and an AI agent?
A workflow answers, retrieves, summarizes or decides: a chat assistant, a retrieval (RAG) system, or a document or decision pipeline. An agent takes actions: it calls tools or APIs on a user’s behalf. Agents take about a week longer to audit, because every tool is another way to fail.
What is monthly testing, and why every month?
After the audit, we test your AI again every month in the test copy of your system, adding new attack and failure techniques as they emerge from our own research and the industry’s, so it is not a re-run of the same tests. We tell you as soon as something fails, and you get a signed quarterly report for your buyers. You don’t want to wait three months to find out your agent has been leaking information. AIUC-1 asks for outside testing at least every three months; we test every month.
Your buyer’s security review
Will our buyer’s security team accept an Autogeny report?
No AI assurance standard is accepted by every buyer yet, so no auditor can promise acceptance. The report is written for the reviewer: what was tested, how, how many times and what failed, with the evidence behind each result. Findings are mapped to the AI questions in the security questionnaire your buyer sends you, the CSA AI Controls Matrix and the OWASP Top 10 for LLM Applications, and we’ll join a call with your buyer’s reviewer if that helps.
Can you help with an AI security questionnaire we’ve received?
Yes. Bring it to a free 30-minute scoping call and we’ll tell you which questions an audit would answer with evidence and which it won’t. After the audit, our findings are mapped to the questionnaire’s AI sections. We don’t write your answers or your policies, because we would then be auditing our own work.
Access and data
Do you need our source code or production access?
We don’t need production access, but we do need read access to the code on your AI’s path: how prompts are built, what its tools can reach and where its output is stored. That isn’t a code security review. We test your system as it runs, with its prompts and tools, on a test copy (your staging system or an isolated copy of it) with test data, never production. We agree access during scoping and work under an NDA and a services agreement.
What happens to our data?
We work on a test copy of your system (your staging system or an isolated copy of it) with test data, never production, and read the code on your AI’s path. Your data stays in isolated test environments for the audit and is deleted afterward unless you start monthly testing. We never use your data to train models or to build tests for other clients.
Independence and fit
Do you fix what you find?
No. The report says what failed and what would fix it, whether that’s a different model or a safeguard in your own system, but we don’t build fixes or write policies for clients we audit. We’ll introduce you to someone who does, and we take no referral fees.
Does Autogeny have a stake in which model we use?
No. We don’t sell models, guardrails, runtime protection or a testing platform, and we take no referral fees, so nothing we find turns into something we sell you.
Who is Autogeny for?
Autogeny is for AI-native companies with roughly $1M to $10M in revenue, with AI already in production, selling to buyers in health, finance, insurance, legal or government who run security reviews. It isn’t for deciding whether to use AI at all, or for low-stakes AI where a wrong answer only costs a re-ask.
Who is behind Autogeny?
Autogeny is an independent AI audit firm in Massachusetts, founded in 2026 by Brad Cunningham. Brad spent 15 years implementing health data systems in over 40 countries, with a focus on data privacy, policy and security. Autogeny is not affiliated with other companies with similar names.
How do we start?
Book a free 30-minute scoping call, or email security@autogeny.ai. Bring the questionnaire if you have one.
Not answered here?
Book a free 30-minute scoping call and ask us directly. If a buyer has sent you an AI security questionnaire, bring it.