There’s no denying AI’s popularity these days. From AI assistants that help solve problems to self-driving cars, AI has become increasingly accessible and widely used.
As more people and companies use AI for everyday functions, it’s not surprising that government agencies like the IRS are using it, too.
Whether it’s a surprise or not, the IRS’s use of AI isn’t new. In August 2022, it reported 10 AI use cases, a number that has since increased to 126 as of June 2025. The IRS is expected to continue implementing AI to improve a range of internal processes and services, including audit selection, fraud detection, and taxpayer service.
To better understand the IRS’s AI operations, the Ways and Means Committee requested a report from the GAO. The U.S. Government Accountability Office (GAO) is a nonpartisan government agency colloquially referred to as the “congressional watchdog.” It reviews government spending to determine if taxpayer money is being spent appropriately and efficiently.
The GAO’s investigation into the IRS’s AI use focused on assessing:
In its report, the GAO notes concerns and challenges that could affect the IRS’ AI initiatives.
The IRS can’t ramp up its AI efforts without adequate staffing to oversee its implementation. However, at the beginning of 2025, the Department of Government Efficiency (DOGE) cut the IRS workforce, intending to reduce federal spending. The IRS staff reduction inevitably hurt its AI development, as the Research, Applied Analytics and Statistics (RAAS) unit, one of the business units most involved in AI, lost about 10% of its workforce. Many of these employees designed, developed, and oversaw AI projects.
Since the staffing reduction, the IRS has rehired some employees, but others found work elsewhere in the interim, leaving the IRS without the level of resources it needs for its AI operations. The GAO report also identified the hiring freeze as a challenge to filling these skills gaps.
To develop AI responsibly, the IRS needs employees with the skills to support the effort. However, the agency has not yet formally discussed the required competencies nor addressed its staffing needs. Instead, the focus is on hiring customer support and audit employees over AI-focused ones. While it can train existing staff to support its AI initiatives, it may not be enough to fill the AI-skill gap, which could jeopardize its ability to maintain the projects currently in development.
The IRS is required to keep an inventory of its current and planned AI use cases and publish a public inventory annually, subject to reporting exceptions. A team is responsible for supporting governance requirements, but the GAO report notes its AI inventory is missing key information. Some entries do not state the AI’s expected benefits, while others omit use-case status details or neglect to list all business units involved. The report even points out that some use cases, namely those that the IRS contracted, like ID.me, the application used to verify taxpayer identity for online IRS accounts, had been missing from the inventory. The report notes that the IRS had begun adding entries for two ID.me use cases.
The report attributes the information shortfall to a lack of quality controls to ensure everything in the inventory is properly documented, as well as missing steps in the IRS’ quality assurance checklist. As for the missing use cases, it points to policy communication issues.
The GAO’s report also looked into the IRS’s management of AI investments. It found that no entity was responsible for managing AI investments across the agency and that the IRS lacked a process to ensure those investments contributed to agency-wide goals.
Taxpayers and practitioners should expect the IRS to continue developing AI. It has already started using AI to answer questions and share information more efficiently through chat and voicebots. AI can also assess compliance and select high-risk tax returns for audit, allowing the IRS to allocate resources more efficiently.
Practitioners should proactively communicate the IRS’s use of AI to taxpayers so they are aware and can set expectations around things like refunds and potential audit selections.
However, AI is not always accurate. While the IRS may find it useful for flagging potential tax returns for audit or investigation, errors can lead to unnecessary and prolonged audits for some taxpayers.
Other concerns about the IRS’ use of AI include potential bias and unfairness, especially when certain tax credits are targeted; privacy concerns about unlawful access to data; and an overall lack of transparency. Practitioners should always advise clients to maintain support documentation. When delivering tax returns, practitioners should warn of potential refund delays and ask taxpayers to forward any IRS communication so they can advise on a course of action.
The report, which was published on March 24, 2026, offers the following eight recommendations to the IRS:
The IRS agreed with all eight recommendations.
AI is transforming processes and operations virtually everywhere and could do the same for the IRS. The agency still has a way to go in its AI efforts, especially in addressing the skill gaps needed to maintain its AI.
By Ashley Akin, CPA
Sources:
About GAO | U.S. Government Accountability Office
Artificial Intelligence: IRS Actions Needed to Address Skills Gaps, Information Quality, and Strategic Management | U.S. Government Accountability Office
Inside the IRS’s Use of Artificial Intelligence | U.S. Government Accountability Office
How The IRS Is Leveraging Artificial Intelligence To Transform Tax Administration | Galleros Robinson Tax Group
Pascrell Letter to Comptroller Dodaro | U.S. House of Representatives Committee On Ways And Means/Subcomittee On Oversight
Disclaimer: The information referenced in Tax School’s blog is accurate at the date of publication. You may contact taxschool@illinois.edu if you have more up-to-date, supported information and we will create an addendum.
University of Illinois Tax School is not responsible for any errors or omissions, or for the results obtained from the use of this information. All information in this site is provided “as is”, with no guarantee of completeness, accuracy, timeliness or of the results obtained from the use of this information. This blog and the information contained herein does not constitute tax client advice.
Join 2,200 of your colleagues and get notified each time a new post is added.
