Many companies describe their use of AI in financial reporting as limited or still in planning. Their auditors are finding something different.
That was the opening message from Christian Peo, Vice Chair of Audit and Assurance at KPMG US, in the Day 2 keynote of the Corporate Financial Reporting Insights (CFRI) Conference. Drawing on nearly 30 years as an auditor, national office partner and former SEC accounting fellow, Peo called AI a revolutionary technology for financial reporting. “If even a fraction of the AI promises come true in financial reporting, it will be revolutionary,” he said. He was equally clear that the change will take time and effort.
AI May Already Be in Your Close
Between walkthroughs earlier in the year and interim testing this fall, KPMG auditors are regularly finding AI in financial reporting processes that management had not identified. Peo described two ways it enters.
The first is by design, which includes service providers. ERP, payroll and valuation vendors are under pressure to add AI features, and those features can arrive through a routine patch that finance never reviewed. The second is individual adoption, where a control operator who once used a spreadsheet starts using an AI model instead.
In one example, a control relied on a spreadsheet lookup with its own checks to keep the data accurate. When auditors returned for the third-quarter instance, the spreadsheet was gone and an AI model had taken its place. The model produced the right answer. ICFR, Peo reminded the audience, also depends on the process and controls that produce that answer, and a change no one identified can surface as a deficiency in January.
A Year-End Checklist
For companies with a December 31 year-end, Peo recommended these steps before the books close:
- Ask about AI use at every level of the organization, including staff-level control operators.
- Look for evidence that has changed or disappeared, such as a reconciliation that used to arrive as a multi-tab spreadsheet.
- Review ERP and other application updates for embedded AI features.
- Check legacy RPA bots and scripts, which are sometimes connected to external models.
- Contact service providers now about whether AI features are active and whether their SOC reports will cover them.
When AI turns up, he advised determining whether it changes the process, the control or the information used in a control. In each case, the questions are the same: can you rely on the output, did it have the right human oversight, and can you evidence that it works for both management and the auditor?
Redesigning the Operating Model
Peo offered a caution for anyone evaluating AI vendors: many products perform well on demo data and fall short on live data. Test on your own data, and confirm the vendor has a SOC report if the tool will support your 404(a) assessment.
The larger lesson concerned process. When an AI tool saves 1% to 2% of hours on a 30,000-hour engagement, KPMG can move a person to other revenue-generating work. On a 500-hour engagement, the same savings amounts to a few hours that cannot be redeployed, so costs rise while savings disappear. KPMG solved this by scheduling groups of smaller engagements together as a single block, and Peo said dozens of similar changes have followed. “Technology wasn’t necessarily the hardest part of that. The operating model was,” he said.
Developing Talent When AI Does the Basics
If AI handles reconciliations, vouching and sampling, how do new professionals build the judgment to become future controllers and accounting managers? Peo said he is more optimistic about that question than he was a year ago.
KPMG rebuilt its intern training around AI fluency: evaluating data inputs and model outputs, challenging assumptions and spotting hallucinations. Interns grasped client workflows significantly faster than under the previous approach, and the firm is shifting first- and second-year assignments toward evaluating outputs and analyzing results. Upfront coaching on core concepts has proven essential. “We are not going to shrink our way into the future,” Peo said, adding that he expects the talent pool companies hire from to arrive better prepared.
What Regulators Are Asking
Peo described the SEC and PCAOB as asking the right questions and engaging directly with firms on new technology. He expects regulators to look less toward new standards and more toward a framework similar to COSO, possibly with changes to audit evidence requirements, since the pressure in using AI centers on evidencing how an answer was reached.
As a starting point, he encouraged attendees to download and study FEI’s AI Framework: Internal Control Over Financial Reporting Whitepaper, developed by FEI’s Committee on Corporate Reporting, calling it an extremely helpful framework and a first step regulators are already reviewing.
In the meantime, the SEC is examining AI disclosures for consistency across earnings releases, earnings calls and periodic filings. Generic claims such as being an “AI leader” need verifiable support. Peo framed strong governance as an accelerator. “The fastest cars have the best braking systems,” he said.
What AI Means for Audit Fees
Peo offered a direct answer to the question he hears at nearly every audit committee meeting. Firms are investing heavily in technology, and a new tool typically adds hours in its first year on an engagement, breaks even in the second and produces savings in the third. With that two- to three-year lag, he said few firms can currently claim higher margins from technology.
Over time, as hours come out consistently and technology costs decline, he expects downward pressure on audit fees, though the size of that change remains to be seen. The near-term benefit shows up elsewhere: companies are spending significantly less of their own people’s time supporting the audit.
Watch Peo’s Keynote On Demand
Registration for CFRI 2026 remains open to FEI Members and Nonmembers. Registrants receive on demand access to Peo’s keynote and the full conference program, along with the CPE-accredited replay, December 1–3, for up to 21 CPE credits.
The program also features keynotes from Mastercard Chief Economist Michelle Meyer and SEC Chairman Paul Atkins, along with updates from the SEC, FASB and PCAOB.
Register for CFRI 2026 »