Dear Acting Commissioner Bessent,
I write to urge the Internal Revenue Service (IRS) to issue comprehensive guidance on the use of artificial intelligence (AI) in tax preparation and compliance. The IRS has repeatedly warned about AI-related chatbots leading to tax fraud risks – including in the “Dirty Dozen” list of scams. Now the lack of transparency and oversight governing AI adoption in the tax profession has created urgent and mounting dangers that require immediate attention through clearer, more detailed regulatory guidance.
The Accelerating Crisis
Artificial intelligence has expanded dramatically in professional services, including tax preparation and compliance. While AI can enhance efficiency in certain routine tasks, its application in complex tax matters presents serious risks that the current regulatory framework does not adequately address. The tax profession faces an inflection point: without swift and comprehensive guidance from the IRS, taxpayers and businesses will face increasing exposure to fraud, penalties, and non-compliance.
AI’s Fundamental Limitations in Tax Matters
The limitations of current AI technology are well-documented and deeply concerning for tax applications:
- Hallucinations Are Mathematically Inevitable: OpenAI has publicly acknowledged that AI hallucinations—instances where AI produces false or fabricated information presented as fact—are mathematically inevitable. Error rates in complex, high-stakes areas like tax law can be as high as 30 percent.1
- 1 “Why Language Models Hallucinate,” OpenAI, September 4, 2025.
- Inability to Handle Complex Circumstances: According to the Taxpayer Advocate, an independent organization within the IRS, AI chat assistants “may encounter difficulties interpreting complex tax laws correctly or considering unique circumstances that could impact a taxpayer’s return.”2
- 2 “Is AI Generated Tax Advice Making the Grade?” Taxpayer Advocate Service, January 7, 2026. The guidance was clear: taxpayers should not solely rely on AI-generated tax advice.
- No Substitute for Professional Judgment: AI cannot replicate the expertise of licensed professionals. For CPAs, accountants, and tax attorneys, professional judgment based on years of training and experience is irreplaceable—especially when dealing with regulated, high-stakes matters.
Emerging Fraud Risks Requiring Immediate Response
We are particularly concerned about accelerating fraud patterns related to AI-driven tax filings, particularly in the research and development (R&D) tax credit space. Already, a growing number of firms are marketing AI-powered solutions promising faster and cheaper R&D tax credits with minimal human oversight. Some are even touting that their solutions can conduct an R&D tax credit study without any actual interviews of the people that are conducting the potential qualified activities. This mirrors the Employee Retention Credit (ERC) fraud scheme that emerged during and after the COVID-19 pandemic—a situation that could have been prevented with earlier guidance—and one which has been featured in your agency’s “Dirty Dozen” list multiple times.
Without clear IRS guidance establishing boundaries on AI use in tax matters, we anticipate:
- Proliferation of aggressive, AI-generated tax claims lacking proper substantiation
- CPAs facing impossible choices between professional liability and competitive pressure to use AI
- Increased audit exposure for businesses relying on inadequately reviewed AI analyses
- Widespread taxpayer harm from fraudulent or non-compliant filings
- Erosion of public confidence in the tax system
Clear Rules of the Road Needed
The IRS has correctly noted “while AI certainly has a role in tax—especially repeatable tasks with predictable outcomes—it cannot replace the need for informed and knowledgeable tax experts.” Yet the tax profession needs more specific, actionable guidance. The American taxpayer and businesses cannot be led to believe that accountability for accurate tax fillings can be outsourced to AI software and tools. There is no gray area for the involvement of AI automation to serve as a mitigation excuse. Errors remain errors, and penalties still apply accordingly.
We respectfully request that the IRS issue comprehensive guidance that:
- Defines Prohibited AI Uses: Clearly prohibit AI from making determinations in complex tax matters, including determining eligibility for credits, substantiating project participation for R&D credits, calculating complex deductions, and interpreting ambiguous tax law provisions.
- Establishes Mandatory Human Oversight Standards: Require that all AI-assisted tax work undergo rigorous, documented human expert review proportional to the complexity and risk level of the matter. Define what constitutes adequate human review for various tax scenarios.
- Requires Documentation and Disclosure: Mandate that CPAs and tax professionals clearly document when, where, and how AI was used in the preparation of any return or submission. Require that this disclosure be visible to taxpayers before filing. Potential examples: Inclusion of the Form 8886 Reportable Transaction form if AI was utilized in the preparation of certain types of tax credits or a Checked box on Form 6765 for Research Incentives that AI was utilized in the preparation for the Credit for Increasing Research Activities.
- Addresses Specific High-Risk Areas: Issue specific guidance on AI use in R&D tax credits, WOTC, and other credits subject to fraud abuse, establishing minimum substantiation and human review requirements.
- Clarifies CPA Liability and Professional Responsibility: Provide guidance clarifying that CPAs cannot outsource professional responsibility to AI and establish standards for when AI use may constitute negligence or fraud.
- Implements Audit and Enforcement Standards: Announce that returns or claims with inadequately supervised AI involvement will be subject to heightened scrutiny, with potential penalties for the tax professional and taxpayer alike.
Protecting the Integrity of the Tax System
The tax system depends on accurate filing, professional integrity, and public trust. As AI capabilities expand, the IRS must expand its guidance to match. Clear boundaries established now will protect honest taxpayers and CPAs from unfair competition with bad actors while reducing future fraud losses and audit burdens.
We stand ready to assist with the development of this guidance and welcome the opportunity to discuss these critical issues further. The time to act is now—before fraud patterns become entrenched in professional services and taxpayer harm multiplies.