The IRS Is Becoming an AI-Driven Counterparty

Listen to this article

Narrated by Charlotte · The Noble House

The disruption does not arrive as a robot taking a tax preparer's chair. It arrives as a notice selected by an algorithm, a client asking an automated assistant for an answer, a filing-season rule change appearing inside software, and an IRS service channel that works quickly for routine cases but leaves difficult ones waiting for human attention.

The shift is real, but the usual description misses its shape. No single federal law presently forces tax-preparation businesses to become accounting firms. Several pressures are converging instead: the IRS is expanding artificial intelligence and advanced analytics; the tax code continues to change rapidly; human service capacity is constrained; professional data-security obligations are rising; and federal advisory bodies are pressing for stronger oversight of non-credentialed preparers.

The profession is dividing rather than disappearing. Routine filing is moving toward automation and price competition. Representation, remediation, planning, judgment, and continuous advisory work are becoming more valuable. A practice owner should audit what to automate, secure what must remain under professional review, and expand only where credentials, partnerships, and demonstrated client demand create real authority.

That is the strategic opening.

The IRS changes the counterparty

The IRS reported 126 active AI use cases as of June 2025. The Government Accountability Office found applications across taxpayer service, operational efficiency, audit selection, compliance work, fraud detection, and criminal-investigation support. It also found an incomplete inventory, missing benefit descriptions for more than one quarter of the reported uses, and weaknesses in strategic management and workforce planning.[1]

This matters because the IRS is not merely experimenting with a public chatbot. AI is entering the machinery that determines which cases receive attention, how anomalies are detected, how work is prioritized, and how routine requests are handled. A tax practice increasingly faces a government counterparty that can process patterns at institutional scale, even when the taxpayer or representative still encounters fragmented human service.

The agency's own governance rules show how consequential these systems have become. The IRS policy effective February 10, 2026 requires inventories, records, testing, privacy controls, and safeguards for high-impact and generative AI. It prohibits unvalidated generative AI from making binding determinations about taxpayer rights without appropriate oversight and from producing taxpayer-facing communications without validation.[2]

Those restrictions do not prove that every IRS deployment is accurate or mature. They prove that the agency recognizes the stakes. The safeguards exist because an automated decision, communication, or prioritization can materially affect a taxpayer.

Federal policy is pushing in the same direction. Office of Management and Budget guidance instructs agencies to accelerate responsible AI adoption while governing high-impact uses and preserving public trust.[11] The operational pressure is therefore unlikely to disappear with one budget cycle. The technologies, controls, and institutional incentives are becoming part of normal federal administration.

A reasonable skeptic could argue that this is ordinary modernization with a fashionable label. Tax software has automated calculations for decades, government agencies have always used risk models, and preparers have repeatedly adapted without abandoning their core business. That objection is serious because it separates durable change from AI marketing. The turn is not that automation exists. It is that automation is spreading across service, case selection, compliance, fraud detection, and professional work at the same time that difficult taxpayers have less reliable access to human resolution.

Compass Decision Intelligence

Analytic module

operational pattern

Two Speed Tax Administration

Routine electronic processing remains strong while complex cases fall into slower human channels.

3 evidence references

Analytic module

strategic inference

Value Migrates Toward Exceptions

Routine filing faces automation pressure; representation and advisory gain value where evidence and professional judgment are required.

3 evidence references

An institutional decision system routing complex taxpayer cases through an automated processing environment. — AI is entering case selection, compliance, service, and fraud detection, changing the institution across the table.

Speed at the center, friction at the edges

The strongest case for automation is also the strongest warning against treating it as a complete substitute for professional service.

During the 2026 filing season, the IRS implemented more than 100 tax-code changes while maintaining strong processing for routine electronic returns. Yet the National Taxpayer Advocate reported deterioration in several channels that matter most when a case is not routine. Live assistors answered fewer calls, voicebots completed only a minority of calls, manual-processing inventory increased, and identity-theft victim assistance remained severely delayed.[3]

This creates a two-speed system. A clean return following an expected pattern may move efficiently. A taxpayer facing an identity problem, inconsistent record, notice, disputed fact, amended return, business question, or enforcement action can fall out of the automated lane and into a slower one. The opportunity lives at that boundary.

That distinction changes the preparer's value proposition. If a practice sells only data entry into a standardized return, software can compress the margin. If it can diagnose why the automated system rejected a case, establish the record, communicate with the agency, and exercise professional judgment, its value rises precisely where automation reaches its limit.

The professional market remains central. The IRS says more than half of taxpayers continue to use a tax professional.[4] Direct File was suspended for the 2026 filing season, so federal filing automation has not replaced professional preparation with a universal public alternative.[4][5]

Tax professionals remain central, but the economic center of their work is moving from form completion toward accountable intervention. That shift calls for deliberate investment, not panic.

Compass Decision Intelligence

Analytic module

control doctrine

Automation With Named Accountability

Use AI to assist intake, research, and drafting while preserving authoritative sources, data safeguards, named reviewers, and escalation.

5 evidence references

Routine cases moving through an automated lane while a complex exception receives human attention. — Routine processing can accelerate even as identity, notice, and exception cases remain trapped outside the fast lane.

Four business models emerge

The first model is the commodity filing shop. Its revenue depends heavily on collecting documents, entering data, producing a return, and repeating the process at seasonal scale. This model remains viable in many communities, but it faces the strongest pressure from consumer software, automated intake, AI-assisted classification, and price-sensitive clients.

The second model is the AI-augmented compliance practice. It uses controlled automation to organize source documents, identify missing information, classify transactions, prepare research, draft routine communications, and manage workflow. The professional remains responsible for verification, judgment, approval, and the final work product.

The third model is the credentialed representation practice. Attorneys, certified public accountants, enrolled agents, and other covered representatives operate under Circular 230 obligations involving competence, diligence, and ethical conduct.[6] Credentials and representation authority matter more when the taxpayer's problem extends beyond preparing the original return.

The fourth model is the continuous advisory practice. It connects filing to bookkeeping, payroll, entity structure, estimated payments, cash-flow planning, compliance monitoring, and business decisions throughout the year. This is the model most closely associated with the investor's concern that tax practices may need to expand toward accounting.

That expansion is not a present federal command. It is a rational response to margin compression in routine work and increasing demand for interpretation, continuity, and intervention.

Tax professionals using secure AI assistance while advising a client in person. — The strongest practices use automation to reduce repetitive work and reinvest professional time in representation, planning, and advisory service.

Law, direction, and prudence are not the same

Strategic planning fails when legal duties, policy recommendations, and commercial forecasts are treated as the same thing.

Some obligations already exist. Covered practitioners must comply with Circular 230.[6] Professional tax preparers are subject to data-security responsibilities, including written safeguards for sensitive taxpayer information.[7][8] IRS privacy guidance restricts how protected information may be introduced into AI systems and warns against unauthorized use of public models.[12] Taxpayer Advocate guidance separately warns that AI-generated tax answers can be inaccurate and must be checked against authoritative law and the client's actual facts.[13]

Other changes remain proposals. The Electronic Tax Administration Advisory Committee and the National Taxpayer Advocate have recommended stronger federal authority, minimum standards, and enforcement tools for non-credentialed preparers.[9][10] These recommendations signal direction and risk. They are not an enacted universal federal credentialing requirement.

The distinction matters. A practice should not spend money responding to a fictional mandate. It should, however, understand that weak security, undocumented AI use, limited representation authority, and dependence on commodity returns can become strategic liabilities even before Congress acts.

What changed in 2026: effects, remedies, and competitive advantage

The new operating environment becomes clearer when the rules are examined where they meet forms, records, payment systems, and automated review. Not every change below is an AI rule. Together, however, they show what happens when new law and new procedures enter an increasingly automated tax-administration system. A practice gains an advantage by controlling the information that reaches that system, recognizing where a case can fall out of the routine lane, and preparing the remedy before the client experiences the failure.

New deductions arrived before source documents caught up

The IRS introduced Schedule 1-A for tax year 2025 so taxpayers could claim new deductions involving qualified tips, overtime compensation, passenger-vehicle loan interest, and seniors.[14] Yet Form W-2, existing Forms 1099, Form 941, other payroll returns, and federal withholding tables remained unchanged for 2025.[15] The benefit therefore exists even when the ordinary document stack does not present all the information needed to determine eligibility or calculate it correctly.

Effect. A preparer who relies only on imported forms can miss a lawful deduction or accept a number that does not satisfy the definitions, phaseouts, filing-status rules, or supporting-record requirements. The information gap shifts work from transcription to reconstruction.

Remedy. Add a targeted intake layer for tips, overtime, qualifying vehicle debt, age, filing status, and modified adjusted gross income. Reconcile employee records with payroll detail instead of assuming that the tax form contains the complete answer. Flag cases requiring employer documentation or professional interpretation before the return enters final review.

Competitive advantage. The practice can find benefits that commodity filing misses while reducing amendment and notice risk. The same intake can support payroll reconciliation, withholding review, tax projections, and employer guidance. One rule change therefore creates a connected service line rather than a one-time form entry.

Form 1099-DA exposes the cost-basis gap

For 2025 transactions, U.S. brokers began reporting digital-asset proceeds on Form 1099-DA. Taxpayers must still report taxable activity even when no form arrives, including activity conducted through foreign brokers.[16] Most 2025 statements do not include cost basis, leaving the taxpayer to calculate it before filing.[17]

Effect. The IRS and the client may receive a gross-proceeds record without the acquisition history needed to calculate the actual gain or loss. Fragmented exchanges, wallets, transfers, fees, and missing basis can make a defensible transaction look like unexplained income or produce an overstated gain.

Remedy. Establish a digital-asset reconciliation workflow before return preparation begins. Collect exchange exports and wallet histories, distinguish transfers from dispositions, reconstruct basis, preserve correction correspondence, and attach a reviewer to unresolved exceptions. The system can organize and match records, but a qualified professional must own the tax treatment.

Competitive advantage. Basis reconstruction, quarterly reconciliation, record maintenance, and notice defense become recurring services. The practice is no longer paid only to report a transaction. It is paid to maintain the evidence that makes the reported result credible.

Electronic refunds make payment readiness part of filing readiness

The Treasury and IRS began phasing out paper refund checks on September 30, 2025. Direct deposit is now the primary delivery method, with alternative electronic methods and limited exceptions for taxpayers who cannot use it.[18]

Effect. A technically correct return may still fail to produce the expected client outcome when bank information is missing or incorrect, the taxpayer is unbanked, or an exception must be documented. The filing decision and the payment-delivery decision remain legally distinct, but the client experiences them as one service.

Remedy. Add a refund-readiness gate before filing. Verify taxpayer-supplied routing and account information through a secure process, require a second review for changes, identify unbanked clients early, and document the alternative-payment or exception path. The practice should verify instructions, never take control of the client's account.

Competitive advantage. Preventing an avoidable refund delay creates immediate, visible value. It also protects vulnerable clients and gives the practice a disciplined way to combine preparation, fraud prevention, and post-filing support.

Automated filters make exception handling a premium capability

The National Taxpayer Advocate reported that more than 14 million individual returns were suspended for additional review during the 2026 filing season. More than one million taxpayers waited beyond normal refund-processing time, with an average delay of about five and a half weeks. IRS employees answered 21 percent of incoming calls overall, while identity-theft victims continued to face resolution periods approaching two years.[3]

Effect. Automation works well for the majority while concentrating hardship among taxpayers whose returns trigger a filter, identity question, missing record, or other exception. Once a case leaves the routine lane, the client needs evidence, status interpretation, and often representation rather than another copy of the return.

Remedy. Classify likely exceptions during intake. Prepare source documents, identity records, prior-year comparisons, and issue-specific evidence as a case packet. Track notices and response windows. Use automation for document organization, deadline monitoring, and grounded drafts, while a named professional verifies the facts and decides when the matter must move to a CPA, enrolled agent, or attorney.

Competitive advantage. A practice that can explain why a case stopped, assemble the record, and guide it toward resolution competes on certainty and relief rather than price. Routine automation and credentialed representation become complementary services: machine speed lowers the cost of preparation, and professional authority resolves the cases that speed alone cannot finish.

These examples identify the practical levers: classification, data completeness, identity assurance, payment readiness, evidentiary quality, and representation authority. The winning practice does not merely add an AI assistant. It joins those controls into one operating system so that information captured during preparation can support planning, client communication, exception recovery, and year-round advisory work. The resulting synergy benefits the client first: fewer missed benefits, fewer preventable delays, stronger records, and a clearer path when the automated system says no.

These cases do not establish that artificial intelligence caused every delay or dispute. They show how statutory complexity, automated matching, payment digitization, and constrained human service combine to reshape the practice. Automation accelerates the ordinary case. It simultaneously increases the value of professionals who can diagnose and resolve the exception.

Four plausible paths from the present

Most likely: automation handles the routine; people handle exceptions

Routine return preparation becomes faster and less differentiated. IRS analytics become more capable at identifying anomalies and prioritizing work. Clients still need professionals, but increasingly for records that do not fit, notices that require explanation, and decisions that cannot be reduced to form completion.

The operating response is to automate intake and repetitive compliance while measuring error rates and preserving human approval. Practices that reduce routine cost without sacrificing trust can defend margins and redirect staff toward difficult cases.

Strategic opening: become the trusted intelligence practice

A practice combines tax knowledge, accounting continuity, secure automation, and client-specific judgment. It becomes the place that understands not only how to file, but what changed, what the change means, what documentation will matter, and which decisions require a CPA, enrolled agent, or attorney.

This model can produce recurring revenue instead of depending entirely on filing season. It also creates a natural boundary between automated preparation and professional advice.

Watch threat: regulation and security outrun the practice

The practice adopts convenient public AI tools without controlling client data, validating outputs, or recording approvals. A breach, incorrect answer, unauthorized disclosure, or regulatory change exposes the weakness at the same time that non-credentialed competitors face stronger scrutiny.

Unmanaged AI inside a business holding sensitive financial identities is the danger.

Tail risk: automation creates an illusion of authority

Clients and staff begin treating fluent machine output as law. Incorrect answers move through intake, preparation, and communication because nobody owns the verification boundary. At the same time, automated government systems make errors that are difficult to contest because the practice allowed its representation and evidentiary capabilities to erode.

This is a low-quality equilibrium: both sides automate, but the taxpayer loses access to accountable judgment.

Build the practice around responsibility

The Noble House begins by mapping the practice and assigning responsibility before automating work. A chatbot may eventually serve a bounded purpose, but it is not the strategy.

First, assess exposure. Identify which services are repetitive, which decisions are consequential, where sensitive data moves, which tasks require credentials, and where staff already compensate manually for broken processes.

Second, design the transition. Divide work into four categories: automate, assist, require professional approval, or prohibit from AI handling. The purpose is not maximum automation. It is reliable allocation of machine speed and human responsibility.

Third, build a secure operating layer. Controlled AI can support document intake, classification, missing-information detection, source-grounded research, communication drafts, workflow tracking, and quality-control checks. Sensitive taxpayer data should remain inside approved environments with access controls, retention rules, and audit trails.

Fourth, encode human review. Consequential outputs receive a named reviewer, evidence references, approval state, escalation rule, and receipt. The system should make responsibility clearer, not obscure it.

Fifth, strengthen the business model. The assessment can identify realistic paths into planning, bookkeeping, payroll support, entity services, representation, compliance monitoring, or recurring advisory work. Expansion should follow demonstrated client need and available professional authority rather than becoming a vague attempt to sell everything.

Sixth, train the organization. Staff need policies for acceptable AI use, source verification, cybersecurity, incident response, client communication, and the point at which work moves to a CPA, enrolled agent, attorney, or other specialist.

Finally, monitor change. Compass Strategic Intelligence can track IRS automation, tax-policy changes, enforcement patterns, professional regulation, security threats, and technology shifts so the modernization plan remains current.

A human reviewer governing a secure evidence and approval workflow. — The operating principle is not maximum automation. It is clear allocation of machine speed and human responsibility.

Start with one bounded proof

A thirty-day readiness assessment can answer the questions that remain specific to the owner's business.

It would map the practice's service mix, credentials, client population, software, data flows, seasonal bottlenecks, notice volume, security posture, and current use of AI. It would identify the highest-risk process and the highest-value automation candidate without altering production systems.

A subsequent ninety-day pilot would implement one governed workflow. A suitable pilot might focus on document intake, missing-information review, notice triage, source-grounded research, or client communication drafting. Acceptance would require measurable time savings, verified accuracy, security controls, human approval, and a rollback path.

Only after that proof should the practice decide whether to expand the system or its services. The sequence is deliberately simple: diagnose the operating reality, prove one controlled improvement, then scale what survived contact with real work.

The questions that decide the route

The next version of this assessment should be based on the owner's answers:

What percentage of revenue comes from routine individual returns, business returns, bookkeeping, payroll, planning, and representation?
Which credentials and representation rights exist inside the practice today?
Which client problems consume the most professional time after filing?
Which states and industries create the greatest complexity?
Where is taxpayer information stored, transmitted, and introduced into third-party software?
Which AI tools, if any, are staff already using?
What work is being declined because the practice lacks capacity, credentials, or an efficient process?
Which recurring service would clients already pay the practice to provide?

Those answers will reveal whether the priority is cost reduction, risk control, credentialed representation, service expansion, or some combination of the four.

The decision window

The IRS is becoming more automated, but the decisive issue is not whether the machine can fill a form. It is who remains capable of exercising accountable judgment when the form no longer resolves the problem.

Tax practices do not need to imitate the IRS or abandon preparation. They need to separate routine processing from professional responsibility, secure their use of AI, and move closer to the decisions clients cannot safely delegate.

The Noble House equips the tax professional with the intelligence, automation, evidence discipline, and operating controls required to remain valuable as the institution across the table becomes more automated.

Research scope notice

A note for Adriana Jensen

A bespoke Compass Special Report dossier prepared for Adriana Jensen. — Prepared for Adriana Jensen: a private Compass assessment of the transition facing modern tax practices.

This Special Report is an evidence-based strategic assessment prepared from publicly available IRS, GAO, OMB, and Taxpayer Advocate materials. Because it was commissioned without a specific statute, case profile, client mix, operating model, or stated pain point, its recommendations are directional rather than individualized legal, tax, accounting, or cybersecurity advice.

Adriana, if you share the concrete issue you want examined, such as a 2026 rule change, service bottleneck, exposure, client segment, or automation decision, The Noble House can prepare a custom-tailored Compass report that tests the exact facts, identifies the practical levers, and maps a focused response for your practice.

Any implementation should be reviewed with the practice's qualified legal, tax, accounting, and security professionals.

Sources

[1] U.S. Government Accountability Office. “Artificial Intelligence: IRS Actions Needed to Address Skills Gaps, Information Quality, and Strategic Management.” March 24, 2026. https://www.gao.gov/products/gao-26-107522

[2] Internal Revenue Service. “IRS Policy for Artificial Intelligence Governance.” Effective February 10, 2026. https://www.irs.gov/irm/part10/irm_10-024-001r

[3] Taxpayer Advocate Service. “Fiscal Year 2027 Objectives Report to Congress: Filing Season Review.” https://www.taxpayeradvocate.irs.gov/reports/2027-objectives-report-to-congress/filing-season-review-27/

[4] Internal Revenue Service. “IRS opens 2026 filing season.” https://www.irs.gov/newsroom/irs-opens-2026-filing-season

[5] U.S. Government Accountability Office. “Department of the Treasury, Internal Revenue Service: Direct File.” https://www.gao.gov/products/b-337954

[6] Internal Revenue Service. “Office of Professional Responsibility and Circular 230.” https://www.irs.gov/tax-professionals/office-of-professional-responsibility-and-circular-230

[7] Internal Revenue Service. “Protect Your Clients; Protect Yourself.” https://www.irs.gov/tax-professionals/protect-your-clients-protect-yourself

[8] Internal Revenue Service. “Safeguarding Taxpayer Data: A Guide for Your Business.” https://www.irs.gov/pub/irs-pdf/p4557.pdf

[9] Internal Revenue Service. “Electronic Tax Administration Advisory Committee 2026 Annual Report.” https://www.irs.gov/newsroom/electronic-tax-administration-advisory-committee-2026-annual-report-includes-recommendations-to-congress-and-irs

[10] Taxpayer Advocate Service. “National Taxpayer Advocate 2026 Purple Book.” https://www.taxpayeradvocate.irs.gov/reports/2025-annual-report-to-congress/national-taxpayer-advocate-2026-purple-book/

[11] Office of Management and Budget. “M-25-21: Accelerating Federal Use of AI through Innovation, Governance, and Public Trust.” https://www.whitehouse.gov/wp-content/uploads/2025/02/M-25-21-Accelerating-Federal-Use-of-AI-through-Innovation-Governance-and-Public-Trust.pdf

[12] Internal Revenue Service. “Privacy for Artificial Intelligence.” https://www.irs.gov/pub/foia/ig/spder/pgld-10-0426-0005-public.pdf

[13] Taxpayer Advocate Service. “Is AI-generated tax advice making the grade?” https://www.taxpayeradvocate.irs.gov/news/tax-tips/is-ai-generated-tax-advice-making-the-grade/2024/06/

[14] Internal Revenue Service. “IRS published schedule taxpayers will use to claim deductions on no tax on tips, no tax on overtime, no tax on car loans, no tax on seniors.” March 2, 2026. https://www.irs.gov/newsroom/irs-published-schedule-taxpayers-will-use-to-claim-deductions-on-no-tax-on-tips-no-tax-on-overtime-no-tax-on-car-loans-no-tax-on-seniors

[15] Internal Revenue Service. “IRS announces no changes to individual information returns or withholding tables for 2025 under the One, Big, Beautiful Bill Act.” August 7, 2025. https://www.irs.gov/newsroom/irs-announces-no-changes-to-individual-information-returns-or-withholding-tables-for-2025-under-the-one-big-beautiful-bill-act

[16] Internal Revenue Service. “Understanding your Form 1099-DA.” Updated June 28, 2026. https://www.irs.gov/businesses/understanding-your-form-1099-da

[17] Internal Revenue Service. “Reminders for taxpayers about digital assets.” January 28, 2026. https://www.irs.gov/newsroom/reminders-for-taxpayers-about-digital-assets

[18] Internal Revenue Service. “Questions and answers about Executive Order 14247: Modernizing Payments To and From America’s Bank Account.” January 27, 2026. https://www.irs.gov/newsroom/questions-and-answers-about-executive-order-14247-modernizing-payments-to-and-from-americas-bank-account

The IRS Is Becoming an AI-Driven Counterparty