The U.S. Government Accountability Office explains how the Internal Revenue Service is increasingly using AI to tackle one of its biggest challenges: the “tax gap,” or the difference between taxes owed and taxes actually paid. AI models are being used to analyze large volumes of tax return data, helping the IRS identify which returns are most likely to contain errors or underreported income, making enforcement more targeted and efficient.
One major application is in audit selection and compliance monitoring. Instead of relying solely on random sampling or manual review, AI helps the IRS prioritize cases with higher risk of noncompliance, such as incorrect claims or suspicious patterns. This allows the agency to use its limited resources more effectively and focus on cases where additional taxes are most likely owed.
However, the GAO also highlights important risks and limitations. AI systems depend heavily on the quality of the data they are trained on, and biased or incomplete data can lead to unfair outcomes—such as disproportionate audits of certain groups. The GAO has also raised concerns about lack of transparency and documentation in some IRS AI models, which could make it harder to ensure consistency, accountability, and public trust.
Overall, the article emphasizes that while AI can significantly improve tax enforcement and efficiency, it must be implemented responsibly. The GAO recommends strong governance, data quality checks, and oversight frameworks to ensure AI systems are fair, reliable, and aligned with public accountability—highlighting that success depends not just on technology, but on how it is managed.
