The IRS has an opinion about what your AI savings are worth
In June 2026 the IRS Office of Professional Responsibility published what it called introductory guidelines for responsible AI use in federal tax practice. Most of it was uncontroversial. One passage was not.
The guidance said that practitioners using generative AI should pass along the efficiencies gained from the technology through billing practices that reflect reduced research and drafting time, and that practitioners should fairly credit clients for cost reductions associated with AI use. It cited Circular 230's prohibition on unconscionable fees as the hook.
On 8 September 2026 the AICPA went public with its objection. As reported by the Journal of Accountancy, the Institute is seeking clarification from the IRS, arguing that the language does not appear to allow for a value-pricing model and overlooks what AI adoption actually costs a firm.
Eva Simpson, AICPA vice president for Member Value, Tax and Advisory Services, called it an overly simplistic view that ignores software licensing costs, implementation expenses, governance requirements and the investment needed to train professionals to use the tools responsibly. Melanie Lauridsen, vice president for Tax Policy and Advocacy, said the AICPA is working with the IRS on clarifying language and FAQs. AICPA president and CEO Mark Koziel was blunter: the OPR language is not authoritative, and could be overstepping a little bit.
That dispute will get resolved somewhere between a clarifying FAQ and a quiet edit. The underlying problem it exposes will not.
The real issue is not Circular 230. It is the hour.
Strip out the regulatory argument and what remains is arithmetic every firm partner already knows.
If a return took four hours last season and AI-assisted research and drafting take it to two and a half, a firm billing hours has just cut its own revenue on that return by 37% while holding its cost base roughly flat. Nobody at the IRS had to require anything. The billing model does it automatically.
The AICPA's position is that value pricing is the answer, and it is right. Jan Lewis, the AICPA chair, framed it precisely: AI-generated work is valuable to the client, that value has a cost and a benefit, and firms need to bill for the service based on the value provided. Koziel noted that many firms have already moved to value pricing, that some bundle return preparation at no additional charge, and that the advisory direction, tax, retirement and estate planning, is where the Institute has been pointing members for years.
Here is the part nobody in that conversation is responsible for solving. Value pricing only works if the client can perceive the value before the invoice arrives. An hourly invoice explains itself. A value-priced invoice does not. It rests entirely on what the client already believes about the firm's expertise, judgment and standing, and those beliefs are formed long before the engagement letter, in exactly the places most CPA firms have no presence at all.
The opportunity: pricing power is a positioning problem
Firms that have already made this transition share a pattern, and it is not a pricing pattern. It is a visibility pattern.
They publish. Their technical point of view is findable, dated and specific enough to be quotable. When a prospect searches a question the firm has written about, the firm's answer is there, and the credential is established before any human conversation happens. This matters more every quarter as prospects ask AI assistants rather than search engines, because those assistants cite what they can find and verify.
They name their process. A firm that can articulate what it does, why it does it that way and what the client receives is a firm that can price the outcome. A firm that can only describe inputs is stuck describing hours.
They are consistent. One good article does not build the belief that supports a value-priced fee. Twelve months of a clear, technically credible point of view does. We laid out the mechanics of this in our guide to digital marketing for accounting firms, and the AI-specific angle in our look at where US firms actually use AI.
There is also a client-facing advisory opening here that most firms are ignoring. Business clients are being told, by the same vendors, that AI will collapse their own service costs. A firm that can explain to a client why its own fee is not falling in proportion to its research hours is a firm that has just rehearsed the exact conversation that client needs to have with its own customers.
The opportunity cost: the invisible firm gets benchmarked on price
Run the alternative forward.
The OPR language stays roughly as written, or the FAQ softens it slightly. Either way the expectation is now in the water. Clients read about AI in accounting, and a reasonable number of them will ask, directly or by implication, why the fee has not moved.
A firm with no public point of view has one answer available in that moment: a discount. It has no body of published work establishing what it knows, no visible process explaining what the client is buying, and no reason on record why its judgment costs what it costs. So it negotiates, and it negotiates from a position where the only measurable thing on the table is time.
A firm that has spent the year publishing has a different conversation. The client has already read three things it wrote. The value was established before the question was asked.
The AICPA's own member research has put AI at the top of the profession's five-year issue list, which we covered in our piece on the 2026 PCPS survey, and the IRS itself is now running AI-assisted audit selection, which we wrote about in IRS AI audits. The profession's exposure to this question is only going one direction.
Your clients are having the identical argument
There is a detail in this story that most firms will read past, and it is the most commercially useful part of it.
The IRS position, stripped to its logic, is that when technology reduces the time a service takes, the fee should fall with it. That is precisely the argument being made right now to every small business client on your book, by their own customers. The marketing agency is told its fee should drop because AI writes the first draft. The law firm is told the same about document review. The bookkeeping client is told it about reconciliation.
A CPA firm that has worked out its own answer to that question is holding an advisory asset, not a pricing headache. The answer has three parts, and it is the same answer in every industry. The client is not buying hours, it is buying an outcome and the assumption of risk. The technology has a cost base of its own, in licensing, implementation and governance, which does not disappear because the output arrives faster. And the judgment applied to the AI output is the part that carries professional liability, which is exactly the part AI cannot assume.
Firms that can walk a business owner through that reasoning become the adviser that owner calls about pricing generally, which is a far stickier relationship than annual return preparation. It is also a natural subject for the kind of content that earns links and citations, because very few accounting firms are writing it.
The practical move is to turn that reasoning into a short published piece and a client email in the same week, then track what it produces. Firms that cannot see which page produced an enquiry end up guessing, which is what our analytics work is for, and the market-level context sits in our cross-market view of accounting firm digital strategy across the UK, USA and UAE.
What a firm should actually do this quarter
Write the pricing page you have been avoiding. Not a price list. A page that explains what the firm does, how it works and what the client receives, in language a business owner can repeat to a spouse. It is the single most-read page on most professional services sites and the one most firms leave as a contact form. Getting more out of the traffic already arriving is conversion rate optimisation work, and it typically moves enquiry rates further than any amount of new traffic.
Publish a technical point of view every month. Consistency beats brilliance. That is the foundation of our SEO work with CPA firms, and it is what gets a firm cited when a prospect asks an AI assistant for a recommendation.
Fix the site before the traffic arrives. A slow, dated site undoes a strong article in four seconds. Our web design work for accounting firms exists for that reason.
Firms wanting the full market-specific playbook should read our digital marketing guide for CPA firms in the USA and our US digital marketing agency page.
The pricing conversation is coming either way
Whether the OPR clarifies the language or not, every CPA firm in the country is going to be asked to justify a fee that no longer maps to hours. The firms that can do it are the ones the market already knows something about.
Triomatic Marketing builds the visibility that makes value pricing defensible. Message Aria, our AI assistant, on WhatsApp from triomaticmarketing.com, or book a free 15-minute discovery call and we will show you what your firm currently ranks for and what it does not.
FAQs
What did the IRS say about billing for AI-assisted tax work?
Guidelines published by the IRS Office of Professional Responsibility in June 2026 said practitioners using generative AI should pass along efficiencies through billing practices that reflect reduced research and drafting time, and should fairly credit clients for cost reductions associated with AI use. The guidance cites Circular 230's prohibition on unconscionable fees.
Why is the AICPA objecting to the guidance?
The AICPA says the language does not appear to allow for a value-pricing model and overlooks the full economics of AI adoption, including software licensing, implementation costs, governance requirements and training. It is working with the IRS on clarifying language and FAQs.
Is the OPR guidance binding on CPA firms?
AICPA president and CEO Mark Koziel has said the OPR language is not authoritative and could be overstepping. The AICPA has separately published its own guidelines and FAQs on the use of AI in federal tax practice.
How does this affect a firm that still bills by the hour?
AI that compresses research and drafting time reduces billable hours on the same work while the firm's cost base stays broadly flat. Hourly billing passes the saving to the client automatically, with no regulatory requirement involved.
What does value pricing require that hourly billing does not?
A value-priced fee rests on what the client already believes about the firm's expertise and judgment before the invoice arrives. That belief is formed by published work, a clearly articulated process and consistent visibility, rather than by a timesheet.