Everyone’s ordering AI, and they’re all putting it on the CFO’s tab.

Now, those running finance teams and leading communities of finance professionals say the strongest CFOs are setting firm controls on AI spending and making teams prove it’s paying off.

“One heavy user can consume more tokens than a hundred colleagues,” Omar Choucair, CFO of financial close software provider Trintech, told CFO Upside. “Without limits, a full year of budget can be utilized in a month.”

For Chris Argent, founder of finance leadership community GENCFO, the spending question is part of a much broader challenge. He said finance leaders recognize the opportunity to redesign work, but implementing that change is proving harder than agreeing it should happen.

“Many leaders understand this shift but are much less clear on implementation,” Argent told CFO Upside.

Whose Budget Is It Anyway?

Companies have experience centralizing analytics and master-data management, Argent said. But AI adoption is spreading less neatly, making it harder to keep track of spending across the business. “Right now AI is everywhere, which means ‘shadow IT’ costs are everywhere,” he said.

Assigning ownership by department appears to be a starting point, according to Argent, though that structure could change as companies better understand the technology’s costs and capabilities. Proving a return remains difficult when the use cases themselves are still developing.

“AI use cases are still emerging, so we can not lean on ‘tried and tested’ finance transformation outcomes,” he said.

Choucair said CFOs often inherit responsibility for AI spending without a clear agreement on who owns the results. “In many companies the AI budget lands with the CFO by default rather than by design,” he said. “If nobody clearly owns it, nobody really owns the outcome.”

Trintech established an AI task force earlier this year with representatives from every department, with tasks to review usage data weekly and reports to executives every two weeks, Choucair said. The company also puts limits on token spending.

His funding threshold starts with familiar questions for CFOs: What problem does the investment solve, what does it cost and what savings can the company reasonably expect? “If we can’t describe the problem clearly and are unable to diagram the data feeds and estimate the anticipated savings, it typically will not get funded,” he said.

Pending Human Review

Getting a project approved does little to guarantee employees will change how they work. Argent described a global hotel chain with a loyal workforce and high employee net promoter scores whose teams were nevertheless protecting their positions and resisting change. In another company, he said, a regional CFO’s efforts to address process problems ran into resistance from the board itself.

“Even the board sees the CFO’s progressive ambitions as outside the CFO remit, when it is clearly the remit of all modern leaders,” Argent said. A CFO expected to justify an AI investment may also need explicit backing to change the processes determining its return.

At Trintech, rolling out enterprise AI licenses revealed different levels of employee readiness, Choucair noted. “Our people sorted themselves into three groups: eager champions, people worried they lacked the skills and a large middle that needed training,” he said. “That is why we treat building the capability internally as seriously as hiring for it.”

The return calculation also has to account for what happens after AI completes its assigned tasks. Trintech’s accounting team, without surprise, uses its own AI platform for reconciliations and close management. Choucair shared how agents gather evidence and recommend next steps, while employees review the output and then make final decisions.

“The best way to improve AI’s ROI is to design verification into the process,” he said. “When the evidence and audit trail arrive with the output, the financial and operational review moves faster, and the team spends its time on the exceptions instead of rechecking everything.”

Even a confidence score needs scrutiny, as Choucair recommends tracking how often employees overturn findings the system rated highly. “The score should shape how much review an item gets, not whether it gets reviewed,” he said.

Choucair also cautioned against tying compensation directly to AI usage, saying employees should be evaluated on what they accomplish with the time automation frees up. Argent expects verification and exception management to become core finance skills, increasing the value of employees willing to keep learning.

Check, Please: Choucair’s test for whether teams are ready to rely on AI is straightforward: Can they explain its output to their auditors? “If they are unable to, the AI tool isn’t ready for full integration, regardless of what the score says.”

Auditors selling CFOs on AI might want to prepare for a follow-up question about the invoice.

Christina Ho, known for her contrarian views during her tenure on the Public Company Accounting Oversight Board and now chief assurance officer at Oath CPAs, an AI-focused audit firm, has a suggestion for finance chiefs whose auditors say they are investing in AI. In a recent interview with CFO Upside, she offered a simple question to ask: “When is my fee coming down?”

“It just doesn’t make sense for you to say you can get efficient,” Ho said, “and for the fees to not only not come down but keep going up.”

If AI can perform substantive audit work and help CPAs deliver an opinion, Ho argues, CFOs should ask how those efficiencies affect what they pay. She now has a business interest in that argument, helping build an audit model she says can deliver those savings while keeping a human accountable for the final opinion.

Ho said Oath prices its audits 25% below traditional firms, though the savings depend on a different style of work. The firm connects directly to clients’ accounting and banking systems and document repositories to obtain records finance teams would otherwise prepare for auditors.

The firm also tests the entire transaction population rather than relying on sampling, with work following clients’ monthly or quarterly closes instead of concentrating around year-end. Ho said AI now makes that approach feasible, allowing issues to surface earlier.

“We want to do machine checking machine,” Ho said, with a human remaining accountable and integrated into the process.

Checking the Checker

For CFOs evaluating those promises, lower fees leave another big question unresolved: How do they measure audit quality?

Ho said her biggest concerns within Oath’s approach are AI hallucinations and automation bias, along with the tendency to accept a system’s output too readily. As plenty of accounting practitioners and community leaders have noted, a polished response can appear convincing to someone who lacks the expertise or context to recognize a mistake.

That gap changes the conversation finance chiefs should have with their auditors, Ho said. When auditors tout their AI investments, she suggested asking directly: “How do you know the AI output is correct?”

Ho’s own build-versus-buy decisions reflect how closely technology and methodology are connected. Buying software often means accepting the process embedded in it, she said. Oath is building its audit technology because she wants to redesign the methodology from the ground up, though she bought a product for the firm’s quality management system because it was faster and cheaper, with updates handled by the provider.

For CFOs weighing similar decisions, she said, evaluating software means examining the way it expects people to work. Whether that process fits the business deserves as much scrutiny as the price.

The CPA Still Counts

Ho expects the value of human judgment to become clearer as AI becomes more widely available. “At some point, when everybody adopts AI, it’s going to become a commodity,” she said. “The differentiator is going to be human judgment.”

Oath calls its practitioners “accounting engineers,” a role combining client work and audit execution with building technology. Ho said employees who understand an accounting problem can use AI to help develop solutions, even without traditional coding skills.

The title does not change who takes responsibility for the opinion, though. Ho said the engagement partner signs it and, of course, must be a CPA. She also requires the credential for senior accounting engineers.

She sees potential for the technology to change firm ownership, too. When asked whether individual CPAs could establish AI-powered practices on their own, Ho said Oath could conceivably offer its platform to practitioners serving their own clients, provided they used its methodology.

Her broader prediction is that AI-generated information will expand demand for assurance beyond financial statements, creating more work requiring professional judgment.

Developing that judgment becomes its own challenge when technology performs tasks that once trained junior accountants, a shift with implications for Ho’s own hiring pipeline. She wants universities to modernize their curricula and bring in practitioners with relevant technology experience, giving students opportunities to build with AI and work through practical problems alongside technology firms.

Inside firms, she explained, young accountants need autonomy and a review process that helps them understand what happened. At Oath, employees with roughly two years of experience can work directly with clients and build tools, she said, while their work remains subject to review.

The product itself can reinforce learning by prompting employees to check particular items rather than simply accepting a completed output. Automating a task does not eliminate the opportunity to understand it, in her view.

“I don’t think it will ever be the case where you can just push a button and then be done with the audit,” Ho said.

Put It on Paper: Ho also explored the profession’s responsibilities while writing Auditing with Courage: Reimagining Assurance and Building Trust in the Age of AI, scheduled for release early next year. She began the book while still at the PCAOB, reflecting on the challenges she faced there and on how auditors sometimes have to act against their own self-preservation instincts. “That was the reason I wrote the book, because I wanted to articulate a vision for the profession,” she said.

Any CFO with a pulse knows AI is taking over accounting’s grunt work, but the experience needed to take responsibility for its output still doesn’t come with an “automate” button.

That conundrum is creating a training problem for finance and accounting: If software takes over the assignments junior accountants once learned on, how do those employees develop the judgment to catch its mistakes?

For Erik Nakamura, CFO of Daily Journal, which publishes California’s largest legal newspaper and operates court software business Journal Technologies, the rise of AI reinforces the value of accounting experience.

Nakamura began his career in KPMG’s audit and tax practices before moving into corporate finance, including financial reporting through real estate brokerage ZipRealty’s IPO and work on e-commerce software company Nogin’s SPAC process. He has also built finance teams, giving him experience developing accountants and overseeing work subject to public-company scrutiny.

“AI makes accounting knowledge and judgment more valuable because someone still has to determine whether the output is right,” Nakamura said in an interview with CFO Upside. “A polished answer is not necessarily a correct one.”

Jack Castonguay, an associate professor of accounting at Hofstra University, sees the other side of that expectation. He told CFO Upside that students entering the profession need opportunities to develop judgment before employers can reasonably expect them to exercise it.

“With accounting firms adopting and deploying AI across many of the tasks junior accountants would have traditionally performed, firms will need to get creative in getting the accountants the experience they need to be able to challenge and review the AI in a low-risk environment,” Castonguay said.

Learning to Review

Hiring fewer junior accountants also raises a question about how firms will develop experienced ones. Business Insider reported in August 2025 that an internal PwC presentation projected reducing US junior associate hiring by roughly one-third by 2028, with AI and offshoring among the factors. The projection illustrates the challenge Castonguay describes: Firms are cutting the roles where accountants traditionally learned the work, even as they still need people who know enough to review it.

The training question becomes clearer when you look at what AI already handles inside a finance department. At Daily Journal, Nakamura said his finance function has found AI useful in preparatory work around the accounting close, SEC compliance and FP&A. Applications include organizing information and preparing presentations, helping the team reach an initial draft or analysis faster.

The review still requires people who understand the underlying accounting, he said.

“Does the analysis reconcile [with] the underlying data? Is the accounting treatment appropriate? Does an apparent exception reflect a real issue or incomplete information?” Nakamura said. “The time savings do not eliminate review; they change where the effort goes.”

That shift, he noted, has implications for hiring. Nakamura values a CPA background and hands-on accounting experience because they help employees recognize missing context and evaluate results.

“The credential matters, but so does demonstrated judgment,” he said. “I want people who can use technology effectively and know when to challenge it.”

In Castonguay’s view, developing that judgment requires practice. He expects simulations, including mock audits and tax returns, to become a larger part of training as firms rethink the work assigned to junior staff.

“I’ve seen comparisons to how pilots often learn to fly on flight simulators before practicing in a real plane and we will need to do the same with accounting,” he said. “That way they can understand what the AI is doing and know where common issues or errors could occur.”

He sees a role for universities in building that experience before students reach their first jobs. Accounting programs, he said, should work with firms to design simulated assignments students can complete during the semester.

“They understand that they can’t know what the AI did wrong or missed if they don’t know what the correct option is to begin with,” Castonguay said.

Taking the Bumpers Down

Castonguay worries that students who use AI to produce correct answers may mistake that success for an understanding of the accounting principles behind them, leaving gaps in their knowledge that only become apparent when they have to work without the tool.

“Students use AI as a crutch,” he said.

He compares that reliance to bowling with bumper guards, where the assistance can make someone appear more capable than they are when it is removed. Students who expect AI to remain available throughout their careers may see less reason to master the underlying work themselves.

For employers, that raises a practical question about how to distinguish a candidate’s knowledge from the quality of the tools they use.

“To see if they understand you have to give them case studies or questions to solve without AI and see how they perform,” Castonguay said. “This probably should occur at the interview stage and then continue on the job. It should be both a performance measurement and a learning opportunity.”

Asked where employers see the biggest gap between their expectations and graduates’ preparation, Castonguay pointed to critical thinking. He emphasized that the issue predates AI, although he believes the technology is making it worse.

His students nevertheless remain interested in traditional audit and tax careers. He said they anticipate reviewing work AI performs, potentially taking on responsibilities previously associated with more experienced staff.

“The work is changing and it’s going to require them to know more, not less,” he said.

Please Use Responsibly: For Nakamura, faster preparation still leaves finance teams responsible for determining whether the information is complete and the conclusions hold up. “AI can assist with gathering information, preparing drafts and reducing repetitive work, but responsibility for the accounting does not move to the tool,” he said.