Tax Filing Gets AI Scrutiny

The IRS has been using some form of artificial intelligence for decades, but its modern initiatives are a direct response to a massive problem: the "tax gap." This gap, the difference between taxes owed and what's actually paid, is estimated to be a staggering $688 billion per year. The primary culprits are underreporting of income, non-filing of returns, and underpayment of taxes. To combat this, the agency is moving beyond traditional audits, which have a high "no-change" rate, and deploying sophisticated AI models. These tools analyze vast datasets and taxpayer history to spot anomalies and flag high-risk returns with greater precision. This shift is part of the agency's strategic plan for 2025 and beyond, which heavily prioritizes AI and data analytics to enhance enforcement. Several key AI systems are already in operation. The Large Partnership Compliance (LPC) model, for instance, targets complex partnership structures like hedge funds and private equity firms that were previously difficult to audit effectively. In one recent tax year, the LPC model identified 82 high-risk partnership returns for examination, a significant increase from previous years. Another tool, the Line Anomaly Recommender (LAR), scrutinizes corporate returns for unusual relationships between income, deductions, and credits. For individual taxpayers, a machine-learning model now analyzes returns and recommends the top three issues most likely to need adjustment. These AI systems are not static; they learn and evolve, with some models running multiple times a tax year to refine their accuracy. This allows the IRS to identify emerging compliance threats and even consider real-time AI checks during the filing process in the future. Beyond just flagging returns, the IRS is also using AI to improve its internal processes. A new system known as Agentforce was rolled out in late 2025 to help staff in the Office of Chief Counsel, Taxpayer Advocate Services, and the Office of Appeals summarize cases and search documents more efficiently. AI-powered voicebots and chatbots are also being used to handle common taxpayer inquiries. However, the expansion of AI has raised concerns about transparency and potential bias. The IRS has been criticized for the "black box" nature of some of its algorithms, making it difficult for taxpayers to understand why they were selected for an audit. Additionally, watchdog groups like the Government Accountability Office have highlighted the risk of unintentional algorithmic bias, pointing to studies that show Black taxpayers are audited at a significantly higher rate. The agency maintains that its use of AI is aimed at creating a fairer and more consistent enforcement system by reducing subjectivity. While the IRS has not disclosed the technical specifics of its models to prevent taxpayers from gaming the system, it has stated that final decisions are still made by human agents. The goal is to focus resources on the most complex and high-risk cases, rather than burdening compliant taxpayers. Looking ahead, the integration of AI into tax administration is set to deepen. The IRS is exploring ways to use AI to notify taxpayers of potential credits or deductions they may have missed. This technological shift represents a permanent change in the IRS's enforcement strategy, with the ultimate goal of closing the substantial tax gap and modernizing its operations.