Artificial intelligence has become a tempting shortcut for tax questions. Type in your situation, get a confident, well-written answer in seconds — often complete with official-looking code sections and court cases. The problem is that some of those authorities do not exist. AI chatbots can and do invent tax law, and relying on their output without verification is how taxpayers and even professionals land in real trouble.
This is general information, not individual tax advice — the right treatment depends on your specific situation.
The Case That Should Give You Pause
In a 2026 Tax Court matter, Clinco (T.C. Memo. 2026-16), a taxpayer’s brief cited court cases that simply did not exist. Judge Mark V. Holmes described the filing as a "bouillabaisse of case names, reporter citations, and legal propositions" that looked like "something cooked up by AI," and warned that "submitting a brief with fictitious caselaw is a recipe for sanctions." Even after the IRS flagged the fabricated citations, they were repeated. The court stopped short of formal penalties but sent an unmistakable message: using AI in tax work requires rigorous due diligence.
Why AI "Hallucinates" Tax Law
A chatbot does not look up the law — it predicts the most plausible-sounding next words based on patterns in its training data. That makes it fluent, not accurate. When it does not know an answer, it often fills the gap with something that reads like a real citation: a case name, a Section number, a dollar threshold, a percentage. These inventions are called hallucinations, and they are especially dangerous in tax because the output looks authoritative and specific — exactly the qualities that make people trust it.
Where AI Helps — and Where It Is Dangerous
Used well, AI is a genuine productivity tool. It is good at explaining concepts in plain language, summarizing long documents, drafting a first version of a memo or email, and helping you organize the questions you should be asking. Where it becomes dangerous is anything that turns on a precise, verifiable fact: the exact citation for a rule, the current dollar threshold or phase-out, whether a deduction applies to your specific situation, or a position you intend to actually take on a filed return. Those are the moments a confident hallucination costs real money.
How to Use It Safely
Treat AI as a brainstorming assistant, never as your final authority. Verify every citation, number, and threshold against a primary source — the Internal Revenue Code, IRS guidance, or the actual court opinion — before you rely on it. Never file a position based solely on a chatbot’s say-so. And remember that AI does not know the facts of your business, your prior returns, or how a dozen rules interact in your specific case. That judgment is the entire value of a professional.
The VarStan Take
We use modern tools, including AI, to work faster — but every number and position that reaches a client’s return is checked against the real law by a CPA. That is the difference between a tool and an advisor: a tool produces a confident answer; an advisor is accountable for it being right. If you have gotten tax guidance from a chatbot and want to know whether it actually holds up, that is exactly the kind of thing worth a quick second opinion before you act on it.
Related reading: Cybersecurity Basics Every Small Business Needs and Protecting Your Family and Business From AI-Powered Scams.