IRS Audit Rates and Changes Ahead

When you look purely at the statistics, the IRS has audited less than 1% of tax filers in recent years. For instance, in 2021, only about 0.3% of all tax returns were audited, according to the latest data.

Audit probabilities are generally higher for certain taxpayer groups, depending on factors like reported income type and claimed deductions. However, even in these targeted groups, audits usually impact fewer than 10% of individuals, and often less than 1%. So, not everyone has to worry too much.

That said, quite a lot has shifted within the IRS over the last year.

Many employees, particularly those with expertise in enforcement and intricate auditing, have either retired or been laid off. A July 2025 report indicated that slightly more than a quarter of both tax inspectors and offices were affected.

Moreover, much of the enforcement funding that was promised through the Inflation Control Act of 2022 has since been scrapped. There’s even a push from the Trump administration for further cuts to IRS funding in the coming year.

In contrast, the agency has also made strides in modernizing outdated systems and increasing the use of AI across several domains, including enforcement efforts. IRS CEO Frank Bisignano mentioned that AI and advanced analytics can help identify areas at high risk for fraud. His vision includes improving collections that exceed historical standards while ensuring compliance.

So, how might these significant changes impact the odds of someone facing an audit in the future?

That’s somewhat unclear for a few reasons. For starters, there’s the question of whether the technology will be used responsibly. Former IRS Commissioner Danny Wuerfel has pointed this out. Additionally, the current number of trained IRS employees could also influence how accurately auditors can select and evaluate potentially questionable returns identified by AI.

In a perfect world, AI could assist the IRS in moving away from auditing compliant taxpayers, consequently lowering the frequency of “no change” audits—those audits that yield no discrepancies and thus no revenue.

AI could also aid the agency in identifying individuals who may underreport income or engage in tax violations.

Barry Johnson, who previously oversaw data and analytics at the IRS, shared that AI makes it easier to spot noncompliance more effectively.

This technology can analyze tax return patterns at a pace that manual methods or earlier statistical models just can’t match, spotting irregular returns that might indicate either mistakes or fraudulent intent.

Wuerfel likened AI’s utility to having night vision goggles, allowing for more informed decisions about which returns to audit and where tax evasion might be happening.

If applied effectively, contact could expand to more taxpayers who owe money, but Johnson emphasized the need for a balance between AI’s capabilities and human oversight to ensure reliable results.

Responsible use of AI, as he stated, is key. The IRS still needs to combine AI-driven insight with human verification to maintain accuracy.

In particular, AI could enhance the efficiency of “correspondence audits.” These are the standard type of audits initiated by the IRS, as opposed to more thorough and time-intensive field or office audits.

During a correspondence audit, the IRS reaches out with a letter about amounts due (or refunds being withheld) concerning specific issues on the filer’s return for a given year. Taxpayers then either pay the required amount or challenge the notice if they disagree.

Nonetheless, if the IRS intends to expand its correspondence audits, Wuerfel advised that the agency must ensure it has enough trained personnel to field inquiries from affected taxpayers.

This also extends to situations where AI brings to light more complex issues, which may call for on-site or office audits.

Currently, there might be a shortage of seasoned enforcement staff, especially given the loss of expertise over the past year. Johnson noted that even with AI, human experts are still necessary to communicate with taxpayers, comb through records, and determine the accuracy of tax returns.

Another factor complicating audits is the AI technology itself, which is advancing rapidly. In the future, it might become capable of even more than it is today.

Wuerfel remarked that while making proactive decisions, one should consider the IRS’s capabilities not just presently but in the upcoming years. The landscape could drastically change by 2028 and 2029 due to AI’s potential evolution.

It’s worth keeping in mind that the IRS can audit returns for at least three years post-filing, or even longer if there’s suspicion of fraud involved.