What Does OPR Alert 2026-19 Actually Say?
OPR Alert 2026-19 clarifies that existing Circular 230 standards already apply to AI in tax practice. Documentation is now expected if the IRS questions your firm's use.
The IRS Office of Professional Responsibility issued OPR Alert 2026-19 on June 24, its first formal guidance on AI use in tax practice. The alert does not create new rules. It clarifies that existing Circular 230 standards—the professional conduct rules for tax practitioners—already apply to AI. The signal is unmistakable: the IRS expects practitioners to document how they use AI, and audits will now ask for that documentation.
Circular 230 is the regulatory backbone of tax practice. It governs due diligence, competence, fees, and confidentiality. For decades, it applied to how practitioners used other people's research, templates, and professional judgment. OPR Alert 2026-19 extends that framework to AI tools. The IRS is not inventing new compliance burdens. It is formalizing what responsible practitioners have always been required to do.
What Four Obligations Define Responsible AI Use?
OPR Alert 2026-19 identifies four Circular 230 obligations that AI use must satisfy. Understanding each one is the starting point for compliance.
Obligation 1: Due Diligence—Verify Every AI Output
Your firm can use AI to draft a return or research citations. That does not reduce your obligation to verify the result. Circular 230 § 10.22 requires due diligence. The alert makes clear this applies to AI-generated content.
Due diligence means checking the citations, confirming the calculations, and verifying the tax law reasoning applies to this client's facts. If the AI cited a 2018 ruling, confirm it hasn't been superseded. If it calculated estimated tax, spot-check the math. The weak point: most firms don't document this work. OPR will ask, "Who reviewed the output and how do you know it was adequate?" File notation like "reviewed by J.P." is not enough.
Obligation 2: Fee Standards—Cost Savings Must Be Transparent
If AI completes research in 20 minutes instead of two hours, the fee must reflect that saving. Circular 230 § 10.27 and § 10.30 require fees to be reasonable and reflect actual work performed. You cannot bill full rates for AI-assisted work without disclosing the efficiency to the client or reducing hours.
Most firms have no fee structure for AI-assisted work. That is the gap. Your engagement agreement needs language addressing fees when AI is used and how you handle efficiency gains.
Obligation 3: Technological Competence—Know Your Tool's Limits
You don't need to be an AI expert, but you need to know what your AI tool can and cannot do reliably. Circular 230 § 10.35 requires competence in methods you use.
Know three things: What is it designed to do? What are its known limits? When should you override its output? If it researches tax law, know whether its training data is current. If it sometimes invents citations, you need to know that and verify output accordingly. Document the tool's known limits before deploying it on engagements.
Obligation 4: Confidentiality—Client Data Only in Approved Platforms
Client information is protected by Circular 230 § 10.29. Tax returns, SSNs, financial statements, bank records, and payroll data are all confidential. They cannot go into tools without clear security agreements and approval.
Free consumer AI tools like ChatGPT do not meet the confidentiality standard. These tools train on their inputs. Your client's data could be used to improve the AI, accessed by other users, or retained indefinitely. Enterprise AI tools with confidentiality agreements and encryption are different. So are approved platforms that your firm has vetted and contracted with explicitly.
The rule is simple: client data goes only into tools your firm has approved in writing and only after confirming the tool has adequate security and confidentiality agreements. That means most accounting staff cannot paste client documents into ChatGPT, Copilot, or Claude without permission from compliance or management. And even then, only if the platform has been formally approved.
What Documentation Will OPR Ask For?
If the IRS opens an inquiry into your firm's AI use, OPR will ask for specific documentation. Here is what to have ready.
First, engagement records. For each file where AI was used, document which tool was used, for what purpose, and who reviewed the output. A one-line notation—"AI tool used for research, reviewed by J.P. on 6/20"—is better than nothing. A detailed memo explaining what was verified is what OPR wants to see.
Second, fee records. If AI materially reduced the time for a task, show how you handled the fee. Did you reduce hours? Did you disclose the AI use to the client and ask them to approve the full fee? Either approach is defensible. Failing to do either is not.
Third, platform approval. Maintain a list of AI platforms your firm has approved for use and which data types each tool can receive. If ChatGPT is not on the list, staff should not use it for client work. If Claude is approved only for research (not client PII), document that boundary.
Fourth, competence documentation. For the AI tools your firm uses, maintain a brief record of what they are designed to do, any known limits, and when human judgment should override them. This can be as simple as training notes or a one-pager from your compliance function.
Free CPA AI Policy Checklist
Before staff paste client data into AI, check the rules your firm is missing.
Get the free checklist and join the Dispatch for practical AI controls, vendor questions, and client-data safeguards for accounting teams.
Free. No spam. You will also get the Nexairi Dispatch.
What Should You Do This Week?
OPR Alert 2026-19 is not an emergency. It clarifies rules that have always applied. But most firms need to act quickly to avoid a compliance gap.
Action 1: Document your AI review process. If your firm uses AI on engagements, write down how you verify the output. Who checks citations? Who verifies calculations? What does adequate review look like for research AI versus drafting AI? Then apply that standard to every file where AI was used. Retroactive documentation is awkward but defensible. Prospective documentation is mandatory.
Action 2: Audit your fee practices against the AI efficiency standard. Pull a few recent engagements where AI was used. Did you adjust fees to reflect the time AI saved? Did you disclose that fact to the client? If the answer is no and you used AI to materially reduce effort, your fee may not be fully defensible. Adjust prospectively and consider a brief disclosure to affected clients if the fee impact was material.
Action 3: Create a list of approved AI platforms and their boundaries. Write down which AI tools your staff can use and for what. Be explicit: if ChatGPT is banned for client data but allowed for firm research, say that. If Claude is approved only with data encryption enabled, document it. This memo protects your firm and guides your team. It takes two hours to write and could save dozens of hours in an audit.
Why This Matters Now
Tax practitioners have been using AI for client work for two years without formal guidance. OPR Alert 2026-19 represents a regulatory shift from silence to clarity. That shift signals that the IRS expects practitioners to have thought about how they use AI and to have documentation ready if questioned.
The IRS is not moving toward an AI ban. It is moving toward an AI audit standard. The firms that document their processes now will have clean files when that audit comes. The firms that do not will be scrambling to explain decisions made in haste and hope.
Sources
Related Articles on Nexairi
Free Assessment
Is your firm ready for AI?
A 5-minute governance check for CPA firms using ChatGPT, Copilot or AI accounting software. Get your score and your top gaps — free.
Jim Smart is the founder and editor in chief of Nexairi. A Business Intelligence Developer with experience building data systems for Verizon, U.S. Army operations, and enterprise finance teams, Jim spent years turning complex data into decisions that executives could act on — dashboards, forecasting models, and automation pipelines across telecom and government contracting. He founded Nexairi to apply that same clarity to AI: making emerging technology understandable and actionable for the operators, accountants, and business owners who need it most. Jim holds GenAI certifications from the University of South Florida Bellini College of AI and completed Springboard's Data Science Career Track.
