You can join a quarter of Americans in using artificial intelligence to prepare your federal income tax return… but you’ll still be spending your time and/or your money on an accountant to check the robot’s work:
The most effective users of AI understand what it’s good at—and what it’s not. Large language models on their own aren’t skilled with numbers or adept at navigating a complex branching system like the tax code that’s supposed to lead to a single answer. A Loyola University Chicago study in November tested LLMs on a common tax question—how much of a home sale is exempt from the capital gains tax?—and found the bots failed two-thirds of the time. “It makes numbers up,” says Joshua Scott, whose Greensboro, North Carolina, firm specializes in the returns of health-care professionals.
…Several software startups offer AI services for tax preparers but say they put careful limits around what the AI is allowed to do. It can be “dangerous,” says David Yue, chief executive officer of two-year-old Accordance, which helps preparers with research and other tasks. Tax returns contain a lot of numbers, and the IRS doesn’t have much tolerance for errors, says Dave Haase, founder of Juno, another tax software startup. “If you have a 99% accuracy rate,” he says, “you have a mistake on every tax return” [Ben Steverman and Charlie Wells, “People Are Using Clause to Do Their Taxes (But Maybe They Shouldn’t),” Bloomberg, 2026.03.18].
“Not skilled with numbers” and “dangerous”—not the slogans I want to see on my accountant’s window.
Plus, Elon Musk and ChatGPT are not bound by the professional confidentiality that binds your local accountant and thus could hand your W2s and bank statements to the courts:
In a recent interview with podcaster Theo Von, Sam Altman, the CEO of OpenAI, acknowledged the risk of sharing sensitive information with OpenAI’s ChatGPT,1 currently the most used AI tool. Altman recognized the lack of legal protection for sensitive information shared with AI tools, unlike information shared with human professionals like therapists, lawyers, and doctors, who are legally bound by client-confidentiality clauses. He further stated that if a privacy matter ends up in legal proceedings, AI providers could be compelled to hand sensitive information to the authorities—a possibility Altman clearly does not look forward to….
Altman’s statement clearly highlights the risk of sharing sensitive information with AI, given the possibility that the information may one day end up being aired in court. This risk applies to all sensitive matters, including taxation, that one might disclose to AI. Especially in complex tax matters, to receive meaningful advice users often must share highly detailed personal and sensitive information. AI can provide rule-based advice, often for free; however, it is very risky to share detailed information due to the lack of legal protection—information that can also be used against you [Avnish Goyal, “Overreliance on AI for Tax Advice: A Cautionary Perspective,” Tax Executive, 2026.02.27].
…and to the general public:
…highly sensitive and personal information is included in tax returns, including full Social Security numbers, personal information, health information, and financial information. Most generative AI tools specifically caution users from inputting this highly sensitive information as it will be used by the tool to train the algorithms and could be used as output to another query by another person. This means that your highly sensitive, identifiable information could be disclosed to someone else without your knowledge [Linn F. Friedman, “Privacy Tip #478—Intrigued with Using AI to Help with Your Tax Return? Please Think Again,” The National Law Review, 2026.02.05].
If you haven’t filed already, your 1040s are due next Wednesday, April 15.