How is AI used in audits?What's changing in 2026, and what isn't.
Short answer: Auditors use AI and data analytics to test entire populations of transactions instead of small samples, flag unusual journal entries, match documents at scale and read contracts. But AI does not sign the audit opinion. There is still no binding, AI-specific auditing standard in the U.S. Existing standards apply to AI-assisted work, and the AICPA's October 2026 practice aid states plainly that the auditor remains responsible for the audit and the opinion.
Key takeaways
- AI lets auditors analyze 100% of transactions in areas such as journal entries, payments and payroll, rather than a sample.
- Common uses: anomaly detection, duplicate payment checks, document matching, contract review and trend analysis.
- No U.S. standard setter has issued a binding AI-specific auditing standard. Existing, technology-neutral standards govern AI-assisted procedures.
- The PCAOB's technology-assisted analysis amendments (AS 1105 and AS 2301) apply to public company audits for fiscal years beginning on or after December 15, 2025.
- The AICPA's October 2026 practice aid covers both a client's use of AI in financial reporting and an auditor's use of AI tools.
- AI does not replace judgment on estimates, professional skepticism, independence or industry knowledge.
- If your finance team uses AI, expect your auditor to ask about the controls around it.
How is AI used in audits today?
AI in audit usually means a mix of data analytics, machine learning and, increasingly, generative AI applied to a client's accounting data and documents. The most common uses are:
- Journal entry testing. Every entry is scored for risk: posted on weekends or holidays, by unusual users, in round amounts, just below approval limits, or to rarely used accounts.
- Duplicate and suspicious payments. Tools match vendor names, bank details, invoice numbers and amounts to find duplicates or vendors that share details with employees.
- Three-way matching. Invoices, purchase orders and receiving records are matched automatically across the whole population.
- Contract and lease review. AI extracts key terms such as dates, renewal options, payment schedules and performance obligations for revenue and lease accounting.
- Analytics and trends. Results are compared month by month, by location or against peers to highlight what looks out of line.
- Drafting and documentation. Generative AI can help summarize documents and draft workpaper descriptions, which the audit team then reviews.
What does AI change about the audit process?
The biggest shift is from sampling to full-population testing. A traditional audit might test 40 invoices out of 4,000 and extrapolate. With analytics, the auditor can examine every invoice, then focus manual work on the items that look risky.
That changes the audit in practical ways:
- Risk is targeted more precisely. Procedures go where the data shows unusual activity.
- Questions become more specific. Instead of generic requests, you may be asked about particular transactions flagged by analysis.
- Data quality matters more. AI-assisted procedures depend on complete, exportable general ledger data.
- Timing can improve. Some testing can happen during the year rather than all at year-end.
In practice: AI makes an audit more thorough. It doesn't automatically make it shorter, because every flagged item still has to be investigated by a person.
What do the AICPA and PCAOB say about AI in audits?
There is no binding AI-specific auditing standard yet. As of mid-2026, neither the IAASB nor the PCAOB had issued one. Existing standards are technology-neutral, so they govern AI-assisted work by their own terms.
PCAOB (public company audits). Amendments to AS 1105 and AS 2301 on technology-assisted analysis were adopted in 2024 and are effective for fiscal years beginning on or after December 15, 2025. When auditors use technology-assisted analysis in tests of details, they must investigate the items identified to determine whether they indicate misstatements or control deficiencies. Auditors must also evaluate the reliability of the electronic information used.
AICPA (private companies, nonprofits, HOAs). In October 2026 the AICPA issued Artificial Intelligence (AI) Considerations for Financial Statement Auditors. It has two parts:
- When a client uses AI: how AI differs from standard automation, how data quality affects reliability, and risks such as model, bias and explainability issues.
- When the auditor uses AI: firm-level quality management, team competence, engagement planning, and partner responsibility for approving and supervising staff use of AI tools.
The AICPA's chief auditor summed up the principle: "Ultimately, the auditor remains responsible for the audit and the audit opinion."
Also note: "the software flagged it" or "the software cleared it" is not audit evidence on its own. The auditor must understand what the tool did, test whether its inputs were reliable, and apply professional skepticism to its output.
What can't AI do in an audit?
- Judge estimates. Allowances for doubtful accounts, inventory reserves, impairments and HOA reserve adequacy depend on facts and assumptions that need human judgment.
- Exercise skepticism. Asking the controller why a number moved, and noticing when the answer doesn't fit, is still a human job.
- Take responsibility. The opinion is signed by a licensed CPA firm subject to peer review and state board oversight. No tool carries that accountability.
- Guarantee independence. Independence rules apply to the firm and its people, not to software.
- Understand your industry. Restricted grants at a nonprofit, reserve studies at an HOA and percentage-of-completion contracts at a contractor all require specialized knowledge.
Traditional vs. AI-assisted audit at a glance
| Area | Traditional approach | AI-assisted approach | What stays the same |
|---|---|---|---|
| Journal entries | Sample of entries selected by criteria | Every entry scored for risk, high-risk items tested | Auditor investigates and concludes |
| Payments | Sample of disbursements vouched | Full population checked for duplicates and anomalies | Support is still examined |
| Contracts | Key contracts read manually | AI extracts terms from all contracts | Auditor evaluates accounting treatment |
| Analytics | Year-over-year comparisons | Monthly, location and peer trends | Unusual results require explanation |
| Opinion | Signed by the CPA firm | Signed by the CPA firm | Firm is fully responsible |
Three worked examples
Example 1: HOA with 1,200 owner accounts. Analytics compare every owner's assessments billed to cash received and flag accounts with unusual credits or write-offs. The auditor then examines those accounts and the board approvals behind them.
Example 2: Nonprofit with federal grants. AI-assisted matching ties grant expenses to budgets and allowable cost categories across the full year. Items charged to the wrong grant or outside the period are flagged for the auditor to review before testing compliance.
Example 3: Distribution company with 40,000 invoices. Duplicate payment analysis finds three vendors paid twice for the same invoice number. The auditor confirms the duplicates, management recovers the funds, and the finding goes into the management letter.
What if your organization uses AI in its accounting?
If AI touches your financial reporting, for example coding transactions, drafting reconciliations or forecasting reserves, your auditor will ask about it. Expect questions like:
- Which processes use AI, and who reviews the output?
- How do you know the data going into the tool is complete and accurate?
- Who can change the tool's settings, rules or prompts, and is that logged?
- What happens when the tool gets something wrong, and how is it caught?
Document these controls before the audit. It makes fieldwork smoother and protects the organization.
A 5-step check to prepare for an AI-assisted audit
- Make your general ledger exportable in a clean, complete format, ideally the full detail for the year.
- Standardize vendor and customer master data so duplicates and mismatches are easy to resolve.
- Reconcile key accounts monthly so flagged items have quick explanations.
- Document any AI tools you use in accounting and the review controls around them.
- Ask your auditor what data they will request and when, so you can prepare in advance.
Common misconceptions
- "AI will replace auditors." It changes what auditors spend time on. It does not replace the licensed professional who signs the opinion.
- "AI audits are automatically cheaper." Some procedures are faster. Investigating flagged items and documenting conclusions still takes time.
- "If the tool found nothing, we're clean." A tool only analyzes the data it is given. Incomplete data produces incomplete results.
- "A very cheap audit must be using AI." Low price usually reflects skipped procedures, not better technology.
Frequently asked questions
How is AI used in audits?
Auditors use AI and analytics to test entire populations of transactions, flag unusual journal entries, detect duplicate payments, match documents and extract contract terms. A human auditor investigates what the tools flag and reaches the conclusions.
Will AI replace auditors?
No. AI automates parts of the analysis, but professional standards require a licensed auditor to plan the audit, exercise judgment and skepticism, and take responsibility for the opinion.
Is there an AI-specific auditing standard?
Not in the U.S. as of 2026. The PCAOB and AICPA apply existing, technology-neutral standards to AI-assisted work, supported by guidance such as the AICPA's October 2026 practice aid.
What is the PCAOB's technology-assisted analysis amendment?
Amendments to AS 1105 and AS 2301, effective for fiscal years beginning on or after December 15, 2025. They clarify auditor responsibilities when using technology to analyze electronic information, including investigating identified items and evaluating data reliability.
What is the AICPA's AI practice aid?
Artificial Intelligence (AI) Considerations for Financial Statement Auditors, issued in October 2026. It covers assessing risks when a client uses AI and managing an auditor's own use of AI tools.
Does AI make audits cheaper?
Sometimes modestly, mostly through faster testing of large data sets. The bigger benefit is a more thorough audit focused on real risks.
Will my auditor ask about our use of AI?
Yes, if AI is used in your financial reporting or internal controls. Expect questions about data quality, review procedures and who can change the tool.
Is my data safe when auditors use AI tools?
Ask your auditor which tools they use, where data is stored, and how confidentiality is protected. Firms remain bound by professional confidentiality rules whatever tools they use.
Talk to us about your next audit
GreyWood CPA performs audits and reviews for businesses, nonprofits and HOAs nationwide, with fixed, transparent fees and senior-level attention on every engagement.
Request a Scoping Call(804) 630-8140