AI Regulation Tracker  /  For CPAs and Tax Professionals

IRS ETAAC 2026 Report: AI Transparency Rules for Tax Pros

The IRS advisory committee wants the agency to disclose how it uses AI, and the same signal tells tax professionals how to prepare their own practices and clients.

IRS ETAAC 2026 Report: AI Transparency Rules for Tax Pros
The Leveraged Years AI Regulation Tracker

The IRS Electronic Tax Administration Advisory Committee (ETAAC) 2026 Annual Report to Congress, released June 18, 2026, supports IRS use of artificial intelligence but recommends the agency publish a plain-language public AI disclosure dashboard explaining how it uses AI, how it manages the risks, and how taxpayers can challenge an AI-assisted action. ETAAC is an advisory body, so its 18 recommendations are nonbinding and the IRS is not required to adopt them. For tax professionals, the report signals rising client expectations that both the IRS and their preparers explain how AI touches a tax return.

For most of the last decade, the argument about IRS technology was about paper: too much of it, too slowly processed. The 2026 report from the agency's own advisory committee moves the argument somewhere more interesting. It says the IRS should not just use more artificial intelligence, but tell people how it is using it. For tax professionals, that is not a distant policy debate. It is a preview of the questions your clients are about to ask you.

On June 18, 2026, the IRS Electronic Tax Administration Advisory Committee released its 2026 Annual Report to Congress, a document with 18 recommendations across six priority areas. One of those areas is artificial intelligence and human-centered design, and the recommendation inside it is worth reading closely, because it draws a line that applies to your practice as much as it applies to the agency.

What ETAAC actually said about AI

Start with what the committee did not say. ETAAC did not oppose IRS use of AI. The report supports the IRS using AI to improve accuracy, speed up service, and strengthen compliance and security, including fraud detection and identity verification, as Forbes tax writer Kelly Phillips Erb reported in her summary of the report.

The condition ETAAC attached is transparency. If the IRS is going to use AI, the report says, taxpayers should understand how. The specific ask is concrete: the committee recommends the IRS publish and regularly update a public AI disclosure dashboard or webpage, written in plain language, that explains how the IRS is using AI, what the process looked like before AI, how risks are being managed, and how taxpayers can seek help or challenge an AI-assisted action. ETAAC also recommends public town hall style meetings on IRS AI projects, especially those that could affect taxpayer rights or safety.

There is a backstory that makes this more than a wish list. Treasury previously published a public inventory of AI use cases that included the IRS, but that inventory is no longer publicly available as of spring 2026. ETAAC's call for a new public dashboard is a direct response to that gap. The committee is not asking for something new in principle. It is asking the agency to restore visibility that recently disappeared.

Why "nonbinding" does not mean "ignore this"

Be honest about the legal weight here, because it matters. ETAAC is an advisory committee, formed and authorized under the IRS Restructuring and Reform Act of 1998, and its members represent tax professionals, software providers, payroll providers, and state and local administrators. It reports to Congress. It does not write regulations, and the IRS is under no obligation to adopt any of the 18 recommendations. Nothing in this report is a new rule you must comply with.

So why should a busy CPA care? Two reasons. First, ETAAC reports are a reliable leading indicator. The committee has spent years pushing for things like real-time credential validation and better Tax Pro Accounts that later showed up on the IRS roadmap. When the committee names AI transparency a priority, that is a signal about where agency practice and client expectations are heading. Second, and more immediately, the standard ETAAC wants to impose on the IRS is a standard your clients will start applying to you. If the public conversation is about whether the taxman discloses its AI use, the natural next question in a client meeting is how their preparer uses AI on their return.

What this signals for the IRS side of your work

The report gives a clear picture of where the IRS wants to point AI, and knowing that changes how you advise clients and read notices.

The committee explicitly recommends the IRS use AI to improve identity-theft filters and to allow vetted tax pros to help verify taxpayer identities. That recommendation sits on top of a real problem the National Taxpayer Advocate has documented. According to figures ETAAC cites from the NTA, current fraud filters produce a high number of false positives: about 54 percent of cases flagged by the system are later determined to be legitimate, affecting nearly 1.3 million tax returns. The NTA also reports that the average time to resolve identity theft victim assistance cases and release refunds was 676 days as of fiscal year 2024.

Read those two numbers together and the practitioner takeaway is blunt. AI-driven fraud screening is already touching your clients, it is wrong more than half the time it flags someone, and the cost of being wrongly flagged is measured in years, not weeks. When a client's refund stalls, "the system flagged you" is now a likely cause, and a slow one to unwind.

The automation divide your vulnerable clients face

There is a second primary source worth pairing with the ETAAC report. The National Taxpayer Advocate's Objectives Report to Congress for fiscal year 2027 describes a growing divide between taxpayers whose issues can be resolved through automated systems and taxpayers who need individualized human help. The report is candid that the people who most often fall outside automated processing channels are frequently those with the fewest resources, the greatest financial vulnerability, and the greatest need for assistance. You can read the Advocate's framing in the FY2027 Objectives Report.

For a practice, that is a client-selection insight, not just a policy note. The clients most likely to get stuck are the ones a chatbot cannot serve: complex situations, prior-year messes, identity theft victims, non-native English speakers, people without reliable online access. As the IRS leans harder on automation, the human help those clients need increasingly has to come from you. That is a service the IRS is quietly outsourcing to the profession, whether or not it says so.

How to use this in your practice

You do not need to wait for the IRS to publish anything. The most useful move is to hold yourself to the transparency standard ETAAC is asking the agency to meet. Here is a framework you can build this quarter.

1. Write your own one-page AI use disclosure. Mirror ETAAC's four elements for the IRS: what AI tools you use in your practice, what that work looked like before AI, how you check and control the output, and how a client can question or opt out of an AI-assisted step. Keep it in plain language. 2. Define your review-before-reliance rule. Decide, in writing, that no AI output goes on a return or into client advice without a preparer verifying it against source documents or authority. Circular 230 due diligence does not soften because software drafted the first pass. Our [Circular 230 AI checklist for tax practitioners](/ai-regulation-news/circular-230-ai-tax-practitioner-checklist) walks through where the duty bites. 3. Prepare a client script for AI questions. When a client asks whether you used AI, or whether the IRS did, you want a calm, specific answer ready, not an improvised one. Practice explaining the difference between AI drafting a document and AI making a decision. 4. Build an identity-flag playbook. Given the 54 percent false-positive rate, treat a stalled or held refund as a probable filter hit. Know the verification steps, the transcripts to pull, and the timeline to set client expectations, because a 676-day average is a client-relationship problem long before it is resolved. 5. Log your AI use per engagement. Keep a short record of which tools touched which client's work. If disclosure norms tighten, or a client disputes an outcome, that log is the difference between a clear answer and a guess. 6. Flag your automation-divide clients. Identify the clients who cannot self-serve through IRS online tools and price and plan for the extra human hours their cases will need.

A simple decision rule ties it together. For any AI tool in your workflow, ask whether you could explain to a client, in one plain sentence, what it did and how you checked it. If you cannot, you are not ready to disclose it, which means you are not ready to rely on it.

What to watch next

The report is a set of recommendations, so the honest forecast is uncertainty. The IRS may adopt an AI dashboard, adopt part of it, or do nothing, and it is doing all of this under real strain: the report notes the agency shed roughly 25 percent of its workforce between January and May 2025 and saw its fiscal year 2026 discretionary funding cut by about $1.1 billion below the prior year, all while implementing the new rules from the One Big Beautiful Bill Act signed July 4, 2025. Transparency projects compete for money and staff that are both shrinking.

What is not uncertain is the direction of client expectations. AI in tax administration is here, disclosure is becoming the expected norm, and the professionals who can explain their own AI use clearly will earn trust the agency has not yet figured out how to build. If you want to develop that fluency, our [Leveraged CPA and Finance course](/leveraged-cpa-finance) is built for exactly this moment, and the [two-minute quiz](/quiz) will point you to the right starting track.

Frequently Asked Questions

Does the ETAAC report create any new AI rules I have to follow?

No. ETAAC is an advisory committee, and its 2026 report is a set of 18 recommendations to Congress and the IRS. Nothing in it is binding law or a regulation, and the IRS is not required to adopt any of it. Treat it as a signal of direction, not a compliance deadline.

What exactly is ETAAC asking the IRS to disclose about AI?

The committee wants the IRS to publish and regularly update a plain-language public dashboard explaining how it uses AI, what the process looked like before AI, how it manages the risks, and how a taxpayer can seek help or challenge an AI-assisted action. It also recommends public town-hall meetings on IRS AI projects that could affect taxpayer rights or safety.

Why does this matter for my practice if it only targets the IRS?

Because your clients will apply the same standard to you. Once "how does the IRS use AI" is a public question, "how does my preparer use AI on my return" is the next one. Building your own plain-language AI disclosure now puts you ahead of that conversation.

How reliable are the IRS AI fraud filters my clients might hit?

According to figures ETAAC cites from the National Taxpayer Advocate, about 54 percent of cases flagged by current fraud filters are later found legitimate, affecting nearly 1.3 million returns, and the average identity-theft case took 676 days to resolve as of fiscal year 2024. Treat a held refund as a likely false positive and set client expectations accordingly.

What is the "automation divide" and which clients should I worry about?

The National Taxpayer Advocate's FY2027 Objectives Report describes a growing gap between taxpayers whose issues automated systems can resolve and those who need human help, noting that the people outside automated channels often have the fewest resources and greatest need. In practice, that means your complex, vulnerable, and low-access clients will need more of your time as the IRS automates more.

What is the single best thing to do this week?

Write a one-page AI use disclosure for your own practice that mirrors what ETAAC is asking the IRS to publish. It forces you to name your tools, your review controls, and your client opt-out, and it doubles as the script you use when a client asks.

Browse the full AI Regulation News tracker

Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.