CNBC66%

Can your tax preparer use AI without telling you? Some experts say IRS rules aren't clear 12%

By Sarah Agostino0%

8/4/2026, 5:35:02 AM

BS Summary: This article contains 10 faulty reasoning types, including Hasty Generalization, Availability Heuristic, and Recency Bias, with Indoctrination as the most egregious example at 12.7% saturation with 137 hits. Analysis detected 623 faulty-reasoning hits from 1,076 analyzed words, generating a BS Score of 23.4% and a BS Rank of 12% (23,255 of 26,303 articles). This article is better (less manipulative) than 88.40% of the article peer group.

Like many other industries, tax firms and accountants are incorporating artificial intelligence more heavily into their workflows. 
But as adoption grows, some experts say IRS privacy rules haven't kept pace  making it worthwhile for consumers to ask how their tax preparer uses AI . 
In June, the IRS released its first AI-related guidance for tax practitioners, including requirements that they review and verify AI-generated work and that billing should reflect efficiencies gained through AI. 
However, it did not specifically say whether using generative AI to prepare tax returns should be disclosed to clients. 
Existing law dictates when tax practitioners must make disclosures and obtain a client's signed permission  for example, when that professional shares your tax documents with your financial advisor or another third-party expert. 
But it also includes exceptions, which typically have applied to tax software. 
And that, say some experts, is where the ambiguity lies. 
"Our members, as well as the AICPA, are clearly trying to get our arms around this because it's obviously all changing, and more and more folks are using AI in their practices, and they want to do the right thing," said Henry Grzes, lead manager for tax practice and ethics with the American Institute of Certified Public Accountants, which represents the accounting profession. 
A 2024 report from the Thomson Reuters Institute, based on a survey of 330 tax and accounting firm professionals, found that about 25% of respondents reported using public-facing, open-source generative AI tools in their work, and just 9% had used proprietary tax-specific generative AI technology. 
Now, many tax practitioners are using AI frequently, and in a variety of ways, according to a June survey from Blue J, an AI-powered tax-research platform, and CPA.com. 
A majority  60%  use AI for tax research at least weekly, up from 33% in 2025, according to the report, which is based on a survey of more than 1,000 tax professionals. 
The report also shows that 44% of respondents use AI for advisory projects, 40% for tax planning, 39% for compliance research, 36% for document analysis and 35% for drafting. 
Section 7216 guidance may be outdated 
Protecting taxpayers' privacy is governed by Section 7216 of the Internal Revenue Code, which says tax preparers generally cannot share or use a taxpayer's information for purposes other than preparing the return. 
Otherwise, the taxpayer must be given a disclosure detailing how and why the information is being shared, and the client must sign it. 
"The problem with 7216 is that the last formal guidance we've received from the IRS goes back to 2013," Grzes said. 
"So when you think about how the tax practice landscape has changed in those 13 years, we're really hoping for some additional guidance from the IRS, both for 7216 and the use of AI," Grzes said. 
In the AICPA's 2026 response to the IRS' annual request for input on which tax issues should be prioritized in the next year, the group asked for additional guidance related to the use of technology  including AI  in tax preparation. 
It was one of dozens of recommendations made by the AICPA. 
"If you have to have a disclosure to send a tax return to, say, a financial advisor that your client has been with for 20 years, wouldn't you think it's appropriate to do that if you're going to send [their information] to an AI tool that's going to prepare the return?" 
said Joshua Youngblood, founder of The Youngblood Group in Dallas and an IRS enrolled agent. 
He also co-owns an AI tool that provides tax research but does not ingest client data. 
"I would argue that [an AI tool] is really not just like your tax software, because a lot of people are relying on AI to make judgment calls and to tell them what to do," Youngblood said. 
Additionally, not all AI is the same, Grzes said. 
For example, there's a difference between using an AI platform that uses input to train itself and a closed AI platform that does not  a difference that the IRS potentially could address through additional guidance, he said. 
"In tax and accounting, trust is the currency of the profession. 
The output shapes filings and client advice, so AI has to meet the standard practitioners already live by," said CPA Elizabeth Beastrom, president of tax, audit and accounting professionals at Thomson Reuters, in an email. 
The company's tax-preparation software, which includes AI tools, is broadly used by accountants. 
"When an AI platform is used solely to prepare a specific client's return, with that client's data protected and used only for that purpose, it plays a role similar to other software and third-party processing tools practitioners already rely on  and the practitioner remains fully accountable and in the loop," Beastrom said. 
Ask about a tax practitioner's AI 'guardrails' 
Grzes said tax practitioners should err on the side of caution and get a signed disclosure from clients if they are using AI to prepare returns. 
"Because we have no formal guidance yet on this topic, our recommendation would be, be safe, as opposed to finding out, 'uh-oh, I should have gotten this,'" Grzes said. 
"Let's say six months from now you get more concrete guidance from the IRS that is more definitive, and you didn't get it, and then now you're exposed." 
The penalties for knowingly violating the statute include a fine of up to $1,000 or jail time of up to one year, or both. 
Other penalties could be applied depending on the specifics of the situation. 
For non-individual tax returns, tax preparers can include 7216 disclosures in their engagement letter, or memorandum of understanding, that they provide to clients, Grzes said. 
For individual returns, however, a 7216 disclosure must be a separate document. 
For consumers, it may be worth asking a tax practitioner whether AI is involved in preparing a return. 
"AI is becoming more ubiquitous every day," Grzes said. 
"If I was a consumer, I would want to know ... what guardrails does the tax preparer have in place to make sure that their personal information isn't being shared with outside or unintended parties." 
He also said to evaluate the answer they get. 
"If the response is something along the lines of 'don't worry,' 'everything is safe' or something similar, without explaining what those safeguards are, I would be concerned," Grzes said. 
Article reasoning-pattern comparisonThis article: 0.0%Sarah Agostino: 0.0%CNBC: 3.7%Confirmation Bias0.0%This article: 4.2%Sarah Agostino: 2.1%CNBC: 1.8%Anchoring Bias4.2%This article: 5.9%Sarah Agostino: 3.0%CNBC: 2.5%Availability Heuristic5.9%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.9%Representativeness Heuristic0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.4%Hindsight Bias0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 2.3%Overconfidence Bias0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 5.1%Framing Effect0.0%This article: 5.3%Sarah Agostino: 2.6%CNBC: 0.6%Loss Aversion5.3%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.6%Status Quo Bias0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.1%Sunk Cost Effect0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 6.5%Optimism Bias0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 1.7%Pessimism Bias0.0%This article: 4.9%Sarah Agostino: 2.5%CNBC: 5.1%Negativity Bias4.9%This article: 4.9%Sarah Agostino: 2.5%CNBC: 1.3%Self-Serving Bias4.9%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.6%Fundamental Attribution Error0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.1%Actor-Observer Bias0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.5%In-Group Bias0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.4%Out-Group Homogeneity Bias0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 2.3%Halo Effect0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.0%Horn Effect0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.0%Dunning-Kruger Effect0.0%This article: 5.9%Sarah Agostino: 3.0%CNBC: 2.1%Recency Bias5.9%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.3%Primacy Effect0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.1%Blind-Spot Bias0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.2%Ad Hominem0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.1%Straw Man0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 3.8%Appeal to Authority0.0%This article: 4.7%Sarah Agostino: 2.4%CNBC: 0.8%False Dilemma4.7%This article: 2.6%Sarah Agostino: 1.3%CNBC: 0.4%Slippery Slope2.6%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.1%Circular Reasoning0.0%This article: 6.6%Sarah Agostino: 3.3%CNBC: 3.3%Hasty Generalization6.6%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.1%Red Herring0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.5%Bandwagon0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 2.0%Appeal to Emotion0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.6%Begging the Question0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 2.8%Post Hoc (False Cause)0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.1%Tu Quoque0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.3%Burden of Proof0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.1%Appeal to Nature0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.2%Composition/Division0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 1.1%Anecdotal0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.0%No True Scotsman0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 1.3%Ambiguity (Equivocation)0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.0%Gambler’s Fallacy0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.1%Middle Ground0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.0%Personal Incredulity0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.1%Special Pleading0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.0%Genetic Fallacy0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 2.1%Unattributed Quote0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.9%Quote-first Misdirection0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 2.8%Biased Writer Voice0.0%This article: 12.7%Sarah Agostino: 6.4%CNBC: 0.7%Indoctrination12.7%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.5%Politically Left Leaning Bias0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 0.1%Politically Right Leaning Bias0.0%This article: 0.0%Sarah Agostino: 0.0%CNBC: 3.8%Attempt to Sell a Product or S…0.0%

1076 words analyzed.

Speakers

3speakers53%attributed speech505writer words
Selected voice

Joshua Youngblood

85%flagged-word coverage
103 attributed words18% of attributed speech30% writer coverage
0%2.5%5.0%Indoctrination-3.6 ptsWriter: 3.6%Joshua Youngblood: 0.0%0.0%

Attribution is sentence-level. Pattern percentages are calculated only from words assigned to that voice.

Loading…
Loading…

Analysis

Hover over highlighted words in the article to view the associated bias or fallacy analysis.