AI is Already Influencing Enforcement
In March 2026, the Government Accountability Office (GAO) released its report “Artificial Intelligence: IRS Actions Needed to Address Skills Gaps, Information Quality and Strategic Management,” which examined how the Internal Revenue Service (IRS) is using and developing artificial intelligence. The findings confirm something that many practitioners have suspected for a while: AI is no longer a future concept at the IRS. It is already embedded in how the agency operates. According to the report, the IRS had 126 active AI use cases as of June 2025, spanning tax compliance and fraud detection, operational efficiency, and taxpayer services. A “use case” is defined by the IRS as “the application of an AI technique to meet a particular business need, such as to solve a problem or increase operational efficiency.”
Even though the report does not explain how returns are selected for audits, it does confirm that AI is already used to identify returns with a higher likelihood of noncompliance. In other words, the IRS is not just collecting more data but increasingly using that data to decide where to focus its limited enforcement resources.
The use of AI matters because it challenges a long-standing assumption among practitioners that audit selection may be largely random or driven by broad statistical sampling. The reality now is more nuanced. AI allows the IRS to process massive datasets, compare information across multiple sources, and identify patterns that would be difficult to detect manually. Returns are no longer evaluated in isolation. Instead they are evaluated in a broader context. While AI is used to flag potentially noncompliant returns, humans are still reviewing the AI outputs and making the final decisions on which audits to pursue.
Funding and Policy Shifts Affect Implementation
The expansion of AI at the IRS is closely tied to recent funding and policy developments. The Inflation Reduction Act provided significant funding intended to modernize IRS systems, including investments in technology infrastructure, data analytics, and enforcement capabilities. This funding created an opportunity for the IRS to accelerate longdelayed modernization efforts and begin integrating more advanced analytical tools into its operations.
However, funding levels and enforcement priorities have shifted across administrations. More recent policy direction has emphasized efficiency, oversight and measurable outcomes. As a result, some funding has been reallocated, delayed, or subject to additional scrutiny. The IRS experienced workforce reductions through attrition and layoffs affecting both its enforcement and technology operations. The practical effect has been a slower and more uneven rollout of AI capabilities than originally planned. While the GAO report does not focus on political dynamics, it notes the operational effects of these shifts, particularly of staffing, planning and execution. The result is an agency that is simultaneously expanding its technological capabilities while navigating constraints that limit how those capabilities are implemented.
GAO Findings and IRS Response
The GAO’s findings can be understood through three themes: expansion, fragmentation and constraint. The expansion is clear. The IRS has significantly increased its use of AI, with 126 use cases identified across the agency as of June 2025, up from 10 in 2022. These cases are not limited to a single function. They span tax compliance and fraud detection, operational efficiency, and taxpayer support. Importantly, the GAO notes that approximately 61% of these 126 use cases were still in development as of June 2025. The IRS is still expanding its AI capabilities, and practitioners can expect continued AI development by the IRS into the foreseeable future.
However, this expansion is occurring within a fragmented structure. The GAO found that the IRS lacks centralized oversight of its AI initiatives. There is no single entity responsible for managing AI investments across the agency, and development is taking place in separate business units without consistent coordination. This decentralized approach increases the risk of duplication, inefficiency and misalignment with broader strategic goals.
Compounding these challenges are significant constraints. The GAO identified skills gaps in key areas related to AI and data analytics, as well as staffing shortages that could limit the IRS’s ability to fully implement and maintain these systems. Due to staffing cuts, the IRS has lost a significant portion of its employees who were filling specialized AI development roles. As a countermeasure, the IRS has been training some of its remaining workforce in AI to continue its AI development objectives. However, the absence of a comprehensive workforce plan further complicates the picture. In short, the IRS is building increasingly sophisticated tools, but it may not yet have the organizational infrastructure needed to fully support them.
The IRS acknowledged these challenges and agreed with the GAO’s list of recommendations. In its response to the GAO, the IRS indicated that it plans to, or is currently, working to improve governance, enhance internal coordination, and better align AI initiatives with its strategic objectives. When funding becomes available, the IRS intends to hire more data scientists, engineers and governance officials to fill the current skills gaps.
These efforts are important, but they are constrained by limited resources. Practitioners should recognize that the current environment is transitional. AI is already being used, but the systems, processes and governance structures surrounding that use are still evolving.
The Shift from Random Audits to Targeted Enforcement
One of the most significant takeaways from the GAO report is not just that AI is being used, but how it is being used. The report explains that AI systems are analyzing data to identify potential noncompliance, indicating a shift toward more targeted, data-driven selection.
This is a fundamental change in how IRS enforcement operates. Rather than relying primarily on broad selection methods, the IRS is increasingly able to identify returns that deviate from expected patterns or contain inconsistencies across data sources. For practitioners, this means that the question is no longer just whether a tax position is technically correct. It is also whether the return presents a consistent data story.
What This Means for Tax Preparers
This shift has immediate and practical implications for how returns are prepared, reviewed and documented. Consistency across data sources is becoming the driving factor in selecting returns for audits. Returns are being evaluated against thirdparty reporting and other data sources. Discrepancies that might previously have gone unnoticed are now more likely to be identified. This includes mismatches between Forms 1099 and reported income, inconsistencies in K-1 reporting, and differences between f inancial records and tax filings.
A simple example illustrates how this plays out in practice. Consider a self-employed taxpayer who operates a small online retail business. During the year, the taxpayer receives $185,000 in gross receipts reported on Forms 1099-K from payment processors. However, on the tax return, the Schedule C reports $142,000 in gross income, reflecting net deposits after fees, refunds and chargebacks.
From a technical standpoint, this reporting may be correct. However, without a clear reconciliation explaining the difference, the return presents a mismatch between thirdparty reporting and reported income. In a data-driven environment, that mismatch becomes a signal. The issue is not whether the taxpayer is ultimately correct but whether the return clearly demonstrates how the numbers tie together.
This example speaks to the broader point that inconsistency itself can create risk, even when the underlying position is defensible, sometimes requiring additional time and effort to respond to an audit or other IRS inquiry.
AI also makes outliers more visible. AI systems are particularly effective at identifying returns that deviate from expected norms. A taxpayer whose deductions, margins or income patterns differ significantly from similar returns may attract attention. This does not mean that unusual situations are inherently problematic, but it does mean that they are more likely to be scrutinized.
Smaller errors and inconsistencies may also bring greater risk. AI systems are designed to process large volumes of data and identify anomalies at scale. As a result, minor discrepancies that might previously have been overlooked may now be flagged for human review.
Documentation remains essential in this environment. While AI may be used to identify potential issues, human agents still evaluate the underlying facts. The ability to quickly provide clear and well-organized documentation can significantly affect the outcome of an examination.
Another practical implication is the growing importance of internal review processes by tax preparers. As AI-driven enforcement increases the likelihood that the IRS will identify inconsistencies, preparers may need to adopt more structured pre-filing review procedures. This could include formal reconciliation checklists, standardized documentation requirements, and secondary reviews focused specifically on data alignment and technical accuracy.
Client communication may need to evolve. Many inconsistencies originate not from aggressive tax positions but from incomplete or misunderstood information provided by clients. Educating clients on the importance of complete and accurate reporting, particularly with respect to third-party forms such as Forms 1099-K, 1099-NEC, and brokerage statements, can reduce risk before the return is even prepared.
Finally, practitioners may find increasing value in leveraging their own technology tools to mirror, at least in part, the IRS’s data-driven approach. While no publicly available software replicates IRS audit selection models, there are tools that can help identify the same types of discrepancies that AI systems are designed to detect. Tax preparation platforms with built-in diagnostic features, including the review modules in Intuit ProConnect, Drake Tax, or Thomson Reuters UltraTax, can flag income mismatches and missing forms before f iling. Data matching tools that cross-reference client records against Forms 1099, W-2 and K-1 data can reveal inconsistencies that would otherwise go unnoticed until flagged by the IRS. Some firms are also adopting workflow tools that make reconciliation sign-offs a prerequisite for return completion, building a documented review trail as a byproduct of the process. Incorporating these tools into the preparation and review process can help tax preparers proactively address the same types of issues that the IRS’s AI systems are designed to detect.
Taken together, these changes show that the role of the tax preparer is evolving. In addition to preparing accurate returns, practitioners are increasingly responsible for ensuring that those returns are consistent, complete and aligned with all available data.
Improvements and Growing Pains
Practitioners and taxpayers are likely to experience both improvements and challenges as the IRS continues to develop its AI capabilities. On the positive side, enhanced data analysis may lead to more efficient processing and more targeted enforcement. This could reduce the number of unnecessary audits and allow the IRS to focus on areas with a higher likelihood of noncompliance.
On the other hand, the ongoing development of AI systems in the current environment may lead to uneven implementation. Because most of AI use cases are still in development (GAO-26-107522, Highlights, p. 1), some target areas may see more advanced capabilities than others. Practitioners should be prepared for a significant transition period in which enforcement patterns may continue to evolve.
The IRS’s use of AI is part of a broader effort to modernize its operations and improve compliance. This effort is reflected in its longterm strategic planning, which emphasizes technology, data analytics, and more effective enforcement.
While the IRS is currently operating under its strategic plan that was created with the previous administration’s priorities in mind (Internal Revenue Service Inflation Reduction Act Strategic Operating Plan FY 2023-2031), an updated strategic plan covering 2026 – 2030 is expected from the IRS soon, although it has not yet been released at the time of the writing of this column. Readers should monitor the IRS website (irs.gov) for the release of the updated plan. The new strategic plan will likely provide greater insight into how the IRS intends to integrate AI into its operations over the coming years, given the current administration’s priorities.
For now, the key takeaway for tax preparers is that the IRS’s approach to AI is still evolving. Practitioners should expect continued changes as new systems are developed and implemented.
Conclusion
The GAO report clearly indicates the IRS is moving toward a more data-driven approach to enforcement. For practitioners, the most important implication is not simply that AI is being used, but that it is changing what triggers scrutiny. The growing use of AI not only increases enforcement, but it also hones its focus. In a system increasingly driven by data, inconsistencies, not just aggressive positions, are becoming the primary audit risk. That shift has immediate implications. It is no longer enough for a return to be technically correct in isolation. It must also be internally consistent, aligned with third-party reporting, and coherent within the broader data environment. For practitioners, the message is clear: accuracy still matters, but consistency now matters just as much.
About the Authors
Emily D. Cokeley, PhD, CPA is an assistant professor at East Tennessee State University. She can be reached at cokeley@etsu.edu.
Lingting Jiang, PhD, CPA is an assistant professor at East Tennessee State University. She can be reached at jiangl@etsu.edu.
References
United States Government Accountability Office. (March 2026) Artificial Intelligence: IRS Actions Needed to Address Skills Gaps, Information Quality, and Strategic Management. Retrieved from https://www.gao. gov/assets/gao-26-107522.pdf
IRS. Internal Revenue Service Inflation Reduction Act Strategic Operating Plan FY2023-2031. Retrieved from https://www.irs. gov/pub/irs-pdf/p3744.pdf
