The audit regulator is rewriting its five-year plan. AI is the argument it cannot settle.
On Monday morning, the Public Company Accounting Oversight Board met to consider a strategic plan that will steer how every US-listed company’s audit is inspected through 2030. Per Thomson Reuters’ reporting, the draft was shaped by 71 comment letters. It was also shaped by something less orderly: a year of public argument, some of it between the regulator’s own members, about what AI should be allowed to do in an audit and what auditors will have to prove about it.
Strategic plans are usually the least readable documents a regulator produces. This one deserves your attention anyway, for two reasons. The first is who is writing it. In July 2025, PCAOB chair Erica Williams resigned and the SEC named board member George Botic acting chair. In January 2026, the SEC appointed a rebuilt board: chairman Demetrios Logothetis, a retired Ernst & Young partner with 40 years at the firm, alongside Mark Calabria from the OMB, Kyle Hauptman of the National Credit Union Administration, and Steven Laughton, then counsel to a PCAOB board member, following more than thirty years at the Treasury. SEC chairman Paul Atkins said the appointments would usher in what he called a new day at the PCAOB, “one of sensible, efficient oversight of auditors.”Whatever the plan says about AI, it will be executed by a board built around that sentence.
The second reason is timing. The plan lands in the middle of the fastest tooling shift the audit profession has seen, one we have been covering all summer as agents moved into the close, the ledger, and the audit file itself. The referee is redrawing the rulebook while the game speeds up.
The regulator spent a year arguing with itself
The clearest way to see the stakes is to put two speeches side by side.
In September 2025, then-board member Christina Ho told a virtual conference hosted by MindBridge, an AI audit software vendor, that the regulator was the bottleneck. “The PCAOB must emerge out of the shadows and issue standards and guidance that accelerate audit firm adoption of such technologies to improve audit quality,” she said, arguing that a technology-neutral approach “is akin to being an anchor that weighs down innovation.” Her example was concrete: an AI tool that tests 100 percent of journal entries instead of sampling. On that point she is right, and it is the strongest version of the pro-AI case: full-population testing is simply better evidence than a sample, and a regulator that cannot say how it will be evaluated slows down something that would help investors. Ho left the board in early 2026.
Three months later, as Accounting Today reported, acting chair Botic stood in front of the AICPA’s December conference and made the opposite half of the argument. He acknowledged that AI “can already enhance risk assessment and evidence gathering by making it possible to efficiently analyze entire populations in certain scenarios,” then added the warning that has followed the profession since: “a growing dependence on AI has the potential to erode qualities that go to the core of what it means to be an effective auditor,” citing research on what automation does to critical thinking and professional skepticism, and cautioning against placing too much trust in technology outputs.
Both positions are sincere, and both are true. That is exactly why the plan is hard to write. This is also not a new dilemma for the institution: a PCAOB task force had named data analytics and AI its priorities back in 2018. Eight years on, when the new board asked the public this spring what its priorities should be, the AI question was still open-ended:“In what ways should the PCAOB consider deploying technology, including AI, to help further its investor-protection mission?”
What the 71 letters ask for
The comment file reads like a tug-of-war over what kind of regulator should meet the AI era.
The firms want modernization and less friction. Thomson Reuters reports broad industry support for building future PCAOB standards on the international IAASB baseline, a view echoed by SEC chief accountant Kurt Hohl. RSM, as Bloomberg Tax reported, urged inspections to focus on whether auditors “exercised sound judgment in critical areas” rather than “documentation formalities.” PwC urged the board to take the opportunity to “reflect on these developments, challenge the status quo, and prepare for the future.” The US Chamber of Commerce pressed the board to deploy AI responsibly itself.
The other side of the file wants the line held. Senator Elizabeth Warren pressed for standard-setting that addresses current risks, “such as protecting auditor independence,”pointing at private equity’s growing stake in accounting firms. The Federal Housing Finance Agency flagged fraud risks, complex valuations, and management review as the areas that keep failing. Investor advocates asked the board not to lose its founding purpose in the redesign.
Strip the politeness and the two camps are asking the same question from opposite ends: when the software does more of the audit, does oversight get lighter because the machines are consistent, or heavier because the remaining human judgment is now the whole game?
The backdrop the plan cannot wish away
Two facts sit under this debate and give it urgency.
The first is consolidation. At that same December conference, per the same Accounting Today report, the director of the PCAOB’s registration and inspections division said more than 90 significant private-equity-related transactions and firm mergers had occurred since 2020, more than half of them in 2025 alone. Botic’s warning was that short-term investor focus “may, over time, begin to shift firm incentives so that profitability outweighs audit quality.”We wrote about what PE ownership does to accounting firmstwo weeks ago; the regulator is now saying the quiet part in public.
The second is what happens when AI output ships without a human seriously in the loop. We covered KPMG’s own AI report citing 40 fabricated sources out of 45 earlier this month. That was a thought-leadership document, not an audit. The uncomfortable question the strategic plan has to answer is what the inspection regime looks like that catches the same failure inside an audit file before an investor pays for it.
Where we think this lands
What follows is our read, not reporting. A board appointed under the banner of “sensible, efficient oversight of auditors,” chaired by a 40-year Big Four veteran, is unlikely to produce an aggressive new AI rulebook. Expect the plan to lean toward international alignment, streamlined inspections, and encouragement of technology rather than new prescriptive standards. However Monday’s vote went, that direction was set months ago by who was put on the board and what the letters asked for.
But notice what every camp in this file already agrees on, sometimes without meaning to. Ho wanted adoption accelerated so that better evidence reaches the investor. Botic wants the auditor’s skepticism protected from the tool. RSM wants inspectors to weigh judgment over paperwork. All three land on the same square: the thing being regulated, increasingly, is not the procedure. It is the quality of human judgment exercised over machine output, and the proof that it happened.
If that is right, inspections drift toward a new set of questions regardless of what the plan’s text says. Which tools touched this file. Who reviewed what they produced. What the reviewer caught. Who signed. Firms that can answer those cleanly will have easy inspections. Firms that treat AI output as finished work will not, and should not.
What it means for you
If you audit public companies, assume the inspection conversation adds a governance thread: how your firm selects, controls, and reviews its AI tools. Document your review of AI-assisted work the way you document any significant judgment, because the judgment you exercised over the output is becoming the evidence that matters most. RSM’s ask cuts both ways here: an inspection regime focused on judgment rather than formalities has to be shown the judgment.
If you prepare financial statements or sit on an audit committee, your auditor’s AI usage is now a fair planning-meeting question: which tools touch our numbers, who reviews their output, and what happened the last time one was wrong. You are buying assurance from a firm supervised by a regulator in transition; it is reasonable to ask for the controls directly.
If you practice outside the US, the push toward IAASB alignment means this is not a purely American story. The baseline being negotiated in Washington is the same one your national standards increasingly build on.
And if you run a smaller firm, the consolidation numbers above are the context for every “future of the profession”paragraph in the final plan. Whether the regulator treats a PE-owned, AI-run audit market as a risk or as efficiency will shape who your competitors are in five years.
The bottom line
Nobody in this fight, not the accelerationists, not the skeptics, not the firms, argues that AI should replace the auditor’s judgment. The argument is about speed, and about who has to prove what. Which means the regulator has arrived, in its own vocabulary, at the place this newsletter starts from every week: AI can do the work. It cannot sign the work. Botic’s “qualities that go to the core of what it means to be an effective auditor” is a regulator’s way of describing the person who reviews the output, catches what it missed, and puts their name on the result. No strategic plan can automate that person, and every credible version of the next five years makes them more important, not less. Keep proof of your own judgment. That is the one document no plan will ever make optional.
— Footnote
Footnote is an independent publication. It is not professional accounting, tax, or legal advice. Our analysis and opinions are based on the public statements, press releases, comment-letter reporting, and articles linked above; statements attributed to officials, firms, and organizations are their publicly reported views, not findings we audited. The PCAOB’s July 20 meeting outcome was not known at press time and details may evolve. We have no consulting relationships with any firm or organization named in this article.