State Tax Agencies Accelerate AI Adoption to Sharpen Audit Focus and Combat Fraud

State tax departments across the United States are increasingly integrating artificial intelligence (AI) tools to significantly enhance the speed and precision of their audits, as revenue officials confirmed at a conference on Tuesday, October 27, 2026. This strategic shift aims to combat fraud more effectively, improve overall tax compliance, and more efficiently allocate limited agency resources.

During a panel discussion at the Hartman SALT Forum in Nashville, Amanda Hiller, acting commissioner of the New York State Department of Taxation and Finance, underscored the motivation behind this technological pivot. “If there’s tools out there that can help us narrow our focus to the taxpayers that should get our attention, then we want to be using those tools,” Hiller stated, emphasizing the goal of more targeted enforcement. While she expressed skepticism that AI would ever solely determine a taxpayer’s liability, she affirmed its potential to guide and inform audit selections.

The adoption of AI by state agencies mirrors efforts already underway at the federal level. The IRS, for instance, is actively leveraging 43 distinct AI strategies for fraud prevention and detection, according to recent reports. The agency employs AI and machine learning to analyze vast datasets, identify high-risk taxpayers, and detect emerging threats, including sophisticated crypto-related fraud. These tools are crucial for managing the sheer volume and complexity of modern tax fraud, which contributed to a net tax gap of $428 billion in 2024.

For small and mid-sized U.S. businesses, this intensified use of AI by tax authorities signals a new era of scrutiny. While the objective is to reduce the burden on honest taxpayers by minimizing false positives, the reality is that any business flagged by an AI system will face a more rigorous audit. This development underscores the critical importance of meticulous record-keeping and robust internal financial controls. In our experience at C&S Finance Group LLC, proactive engagement with tax compliance and financial risk management has never been more vital. Businesses must ensure their financial data is not only accurate but also consistently presented across all filings, as AI excels at identifying inconsistencies across disparate data sources. We often advise clients that a strong foundation in tax preparation and compliance is their best defense against potential audit triggers in this evolving landscape. For guidance on navigating these changes, businesses can reach out to C&S Finance Group LLC at csfinancegroup.com.

AI’s capabilities extend beyond basic data analysis. It combines traditional machine learning on financial data with large language model (LLM) analysis of returns and supporting documents. This two-stage approach allows AI models to flag statistically unusual patterns, after which LLMs examine flagged returns to assess the likelihood of actual fraud versus legitimate but unusual circumstances. This method has been shown to reduce false positive rates by 40-60% compared to using machine learning alone, thereby saving audit resources and reducing unnecessary burdens on compliant taxpayers. Furthermore, AI excels at integrating multiple data sources, detecting cross-system fraud schemes that human analysts would struggle to uncover efficiently.

Beyond tax fraud, AI is also being deployed to detect benefit abuse and procurement irregularities. For benefit fraud, AI analyzes application patterns, cross-references data, and identifies synthetic identities. In procurement, it monitors bid patterns, vendor relationships, and contract modifications for anomalies. These applications highlight AI’s versatility in combating various forms of public sector fraud, which globally exceeds $500 billion annually.

Despite the clear benefits, agencies deploying AI for fraud detection are acutely aware of the need to balance effectiveness with citizen rights. Essential safeguards are being implemented, including privacy impact assessments before deployment, bias audits to ensure AI does not disproportionately flag certain demographics, and human review of all AI-flagged cases before any enforcement action is taken. Transparent appeal processes are also crucial. For the protection of sensitive taxpayer data, agencies are opting for on-premises deployment of LLMs like Llama 4 or Mistral Large 3, rather than cloud-based solutions.

The strategic shift towards AI-powered audits by state tax agencies represents a significant evolution in tax enforcement. For small and mid-sized businesses, this means an increased likelihood of targeted audits based on sophisticated data analysis. The emphasis is now firmly on data integrity and consistency, as AI systems are designed to identify even subtle discrepancies that might have gone unnoticed by traditional audit methods. This technological advancement, while aimed at fairness and efficiency, demands a heightened level of preparedness from all taxpayers.

As state tax departments continue to refine their AI strategies, businesses should anticipate an ongoing increase in the sophistication of audit selection. Monitoring legislative developments and agency announcements regarding AI implementation will be crucial for understanding the evolving landscape of tax compliance and enforcement in the coming years.