TLDR: AI in IRS Auditing

• The IRS is using artificial intelligence (AI) and other improved technology to aid in its audit processes.
• Concerns exist about the potential privacy, bias, transparency, and unintended consequences of AI in audits.

This tax filing season, the IRS is implementing the use of artificial intelligence (AI) and other improved technology with funding from the Inflation Reduction Act of 2022 in its audit processes to help compliance teams better detect tax avoidance issues in certain areas where audit coverage has declined, including large partnerships, large corporations, and employment tax returns.

Although there can be benefits to utilizing AI for auditing taxpayers, there are also concerns about what guardrails are in place regarding privacy, bias, transparency, and the potential for unintended consequences, such as taxpayers being subjected to audits for returns simply falling outside the usual fact patterns.

Per the IRS’s Strategic Operating Plan for fiscal year (FY) 2023 through 2031, expanded enforcement is on the radar for large partnerships and large corporations, with additional focus on the employment tax returns these and other entities file. The IRS said it will improve tools and processes for auditing these returns and “pursue noncompliance through a variety of robust mechanisms, including audits and non-audit contacts.” However, there is some uneasiness about the use of AI in IRS audits by not realizing the full picture of a tax return.

James Creech, a senior manager with Baker Tilly’s tax advocacy and controversy team, expressed concern about the potential consequences of AI audits. He questioned what would happen if the IRS AI identifies an issue that falls outside the norm. Creech also raised concerns about the human element of AI in IRS audits, explaining that humans may interpret AI findings and enforcement in different ways. Additionally, he discussed the potential impact on tax practitioners and whether their guidance to clients would change in response to AI implementation.

Creech also discussed the use of AI in examining claims for the employee retention credit (ERC). Due to the limited amount of data available for ERC claims, he suggested that AI may not be effective in these audits. He also shared advice for taxpayers who believe they have valid ERC claims and are concerned about an audit. He emphasized the importance of organizing payroll data and calculations, and staying updated on IRS audit notices.

Looking toward the future, Creech acknowledged that the attitude within the IRS about the ERC is changing. He explained that during the COVID-19 pandemic, the IRS was more focused on helping taxpayers, but that attitude has shifted to a more skeptical approach. He also highlighted a potential legal challenge to the IRS’s guidance on the ERC, which could lead to changes in the guidance and may not be favorable to taxpayers.

Overall, the use of AI in IRS audits presents both advantages and concerns. While AI technology can improve the efficiency and effectiveness of audits, there are potential risks in terms of privacy, bias, transparency, and unintended consequences. The human interpretation of AI findings and enforcement is another important aspect to consider. For tax practitioners and taxpayers, there may be adjustments in how they approach audits and provide guidance to avoid potential issues with AI algorithms.