Policy
IRS Bulletin 2026-19 Sets New AI Compliance Rules for Tax Pros
The Office of Professional Responsibility now requires tax practitioners to demonstrate competence, verification, and human oversight when using generative AI tools.
Key takeaways
- IRS Bulletin 2026-19 makes AI competence and oversight mandatory under Circular 230, applying existing professional responsibility rules to AI-enabled tax workflows.
- Tax practitioners must establish documented verification processes proving human judgment remains in control of AI-generated content and conclusions.
- The IRS cites real sanctions against attorneys and firms that filed fabricated AI-generated citations or delivered reports with invented content.
- Firms must understand AI limitations including hallucinations, embedded bias, data security risks, and confidentiality vulnerabilities before deployment.
- Competitive advantage comes from pairing AI adoption with strong controls and documented compliance processes, not from avoiding the technology.
IRS establishes mandatory AI standards for tax practice
The IRS Office of Professional Responsibility issued Bulletin 2026-19 in June 2026, creating the first explicit regulatory framework governing artificial intelligence use in federal tax practice. The guidance applies existing Circular 230 requirements—specifically sections 10.35, 10.36, and 10.37—directly to AI-enabled workflows, making technological competence and proper AI oversight mandatory rather than optional.
The bulletin arrives as generative AI tools have become commonplace in tax and accounting firms, from research platforms to document review systems. Unlike traditional AI that performs defined tasks, generative AI creates original content by predicting patterns from training data—a capability that introduces distinct professional risks.
Why it matters
Bulletin 2026-19 transforms AI adoption from a competitive choice into a compliance obligation. Tax practitioners must now demonstrate documented verification processes, maintain human oversight of AI outputs, and show competence in understanding what AI tools can and cannot reliably do. Firms that fail to meet these standards face the same consequences as any Circular 230 violation: professional discipline, court sanctions, reputational damage, and potential client notifications. The regulatory baseline has shifted, and practitioners can no longer treat AI as a neutral productivity tool without accountability measures.
Real consequences documented
The IRS bulletin cites specific cases where improper AI reliance led to professional sanctions. In 2023 and 2024, courts sanctioned multiple attorneys for filing briefs containing fabricated citations generated by AI. In 2025, Deloitte Australia issued public apologies and refunded fees after delivering a government report filled with invented quotes and references reportedly produced by AI systems.
These examples illustrate the core risks the IRS identifies: hallucinations (fabricated citations and false statistics presented confidently), embedded bias from training data, lack of transparency in how conclusions are reached, data security vulnerabilities, and confidentiality breaches when client information influences unrelated responses.
What compliance requires
Bulletin 2026-19 establishes three core requirements for tax practitioners using AI. First, practitioners must understand the capabilities and limitations of the AI tools they deploy. Second, firms must establish documented, firmwide procedures demonstrating Circular 230 compliance. Third, practices must maintain verification processes proving that professional judgment remains with human practitioners, not algorithms.
The bulletin explicitly states that professional responsibility does not transfer from practitioner to technology. AI-generated content must be treated as a starting point requiring verification, not a finished product. Firms must choose enterprise-grade solutions with appropriate security controls and establish clear protocols for staff use.
Competitive advantage through responsible adoption
The guidance does not discourage AI adoption—it requires responsible implementation. According to Thomson Reuters Director of Tax Compliance Eric Clements, firms succeeding with AI share common approaches: they recognize AI as a tool that amplifies expertise rather than replaces it, they vet solutions for security and reliability, and they maintain rigorous verification processes.
Firms that pair AI adoption with strong controls and documented oversight will gain competitive advantages while meeting regulatory expectations. The distinction between compliant and non-compliant practices now centers on demonstrable processes, not technology choices.
These details were first reported by Thomson Reuters Tax & Accounting in an analysis by Eric Clements.
