Workforce Reductions at the IRS: A Growing Concern
The effects of President Trump Administration’s federal cuts on government spending continues to raise concerns about the enforcement of federal tax laws. Due to President Trump’s series of executive orders and Office of Personnel Management orders relating to reductions in workforce and deferred resignation program, the number of federal employees in the workplace of the Internal Revenue Service (IRS) has already been reduced by 11% with more downsizing anticipated now that the tax filing season has concluded. See the Treasury Inspector General for Tax Administration Report No. 2025-IE-R017. Despite Treasury Secretary Scott Bessent, who recently appeared before the House Appropriations Committee telling House lawmakers that better IT and artificial intelligence will “enhance collections” at the tax agency, as he believes an “AI boom” will counteract staff cuts, it is worrisome to rely on AI to evaluate tax issues that necessitate human interpretation.
Can AI Fill the Gap Left by Human Expertise?
Federal tax laws are not just numbers. It requires more. Many areas in tax law have qualitative subjective components involving interpretation of facts and circumstances, not just scanning for certain buzzwords. For example, the federal research tax credit requires a subjective analysis of a four-part test to interpret if a person’s activities are considered qualified activities to be eligible to claim the credit.
The ERTC Denial Problem
We have also seen far too many IRS denial letters for credit claim amounts where notice letters are automated and sent to taxpayers without providing specific reasoning. For example, let’s take a look at the processing (or lack thereof) of the Employee Retention Tax Credit (“ERTC”), a refundable tax credit for certain eligible business employers and tax-exempt organizations that had employees and were affected during the COVID-19 pandemic. The requirements are different depending on the time period for which an employer claimed the credit. Nevertheless, it didn’t appear that the IRS spent time evaluating many of these claims but rather issued significant amounts of automated denial letters. Many of these denial letters simply indicated that a taxpayer is not eligible for ERTC either because they did not suffer a significant decline in gross receipts or were not subject to a government shutdown during a tax quarter. From our experience what we have seen is that many of these denial letters contain inaccuracies as the IRS doesn’t have the requisite quarterly breakdown of the gross receipts or doesn’t take into consideration their own published guidance regarding the alternative election to qualify a 2021 tax quarter. The IRS has acknowledged errors in these notices. As mentioned above, simply sending automated letters using AI tools is not the solution as in many cases, ERTC qualification, requires human interpretation and analysis to review and make a determination.
Although AI may be used as a tool to help with enforcement of federal tax laws, it is by no means a substitute for human analysis, insight and interpretation. The drastic cut of federal employees within the IRS cannot simply be replaced with AI and expect that collections and enforcement of federal tax laws will simply thrive. The long term effect of losing these federal employees may not truly be known; nevertheless, it seems likely that it will impact the IRS’ ability to provide timely and effective taxpayer services as the need for human evaluation appears prominent in various federal tax laws.
