How the IRS Is Using AI to Target High-Risk Tax Returns?

The IRS is actively integrating artificial intelligence and machine learning into its audit selection system. This marks a clear move away from traditional methods that relied heavily on outdated statistical models and manual reviews. Those earlier systems often produced high “no-change” audit rates, where audits resulted in no additional tax liability.

With an estimated $688 billion annual tax gap, the IRS is now focusing on precision-driven enforcement. Reports from the Treasury Inspector General for Tax Administration highlight that while AI models show strong potential, they still require better feedback loops, evaluation systems, and ensemble learning to reach full efficiency. High-income individuals, large partnerships, and corporations are expected to face increased scrutiny under this evolving system.

How AI Is Transforming Audit Selection?

The IRS now uses advanced AI models across multiple taxpayer categories. These systems focus on pattern recognition and relational analysis instead of isolated data points. The objective is clear—reduce unnecessary audits and improve detection accuracy.

·        Individual Taxpayers (Form 1040)

For individuals, machine learning models analyze each return and identify the top three areas most likely to need adjustment. These models have been in development even before federal AI governance policies were introduced. The system prioritizes risk signals instead of random sampling.

·        Mid-Sized Corporations (Form 1120)

Corporations with assets between $10 million and $250 million are now evaluated using the Line Anomaly Recommender (LAR). This system replaces the older Discriminant Analysis System.

LAR examines relationships between financial elements such as income, deductions, and credits. Early results show fewer no-change audits and broader coverage across returns.

·        Large Partnerships (Form 1065)

The IRS has introduced the Large Partnership Compliance (LPC) model for complex structures like hedge funds and private equity firms. This system combines machine learning with expert human review.

It has already identified high-risk returns that were previously overlooked. This marks a significant improvement in enforcement for partnership structures.

Common AI Audit Triggers

AI models are trained to identify inconsistencies and unusual patterns. Some key red flags include:

• Low reported income paired with high-value assets
• Multi-year inconsistencies in financial reporting
• Complex layered partnership structures
• Mismatched deductions and income across related entities
• Unusual ratios between income and deductions

These signals help the system detect potential noncompliance with greater accuracy.

Expanded Targeting in 2025

The scope of AI-driven audits is expanding beyond large entities. The IRS may now also flag:

• High Schedule C losses from small businesses
• Disproportionate charitable deductions
• Unreported cryptocurrency transactions
• Rounded-off expense entries
• Unsupported tax credit claims

Freelancers, gig workers, and self-filers face higher exposure. Even minor reporting errors can trigger automated review.

Strategic Implications for Taxpayers

The adoption of AI is a long-term enforcement shift. It is not a temporary upgrade. Taxpayers must adapt by improving accuracy, maintaining detailed records, and ensuring consistency across filings. In complex cases, working with a criminal tax attorney can help manage risk and respond to high-stakes audits effectively.

Similarly, businesses dealing with large-scale compliance issues often benefit from consulting top IRS attorneys who understand how these AI-driven systems operate.

Final Takeaway

AI is making IRS audits more targeted and data-driven. The focus has shifted from volume to precision. Taxpayers should act early, strengthen compliance frameworks, and eliminate inconsistencies before they trigger automated flags.