AI in Tax Is Moving Fast. Are We Ignoring the Risks? | Danny Werfel

AI in tax is advancing at a staggering pace, but are tax professionals moving too fast to understand the risks?

In this episode of the AI Native Accounting podcast, Kacee Johnson sat down with Danny Werfel, former Commissioner of the Internal Revenue Service and current Executive in Residence at the Johns Hopkins University School of Government and Policy, to talk about what responsible AI adoption should look like across the tax profession.

Werfel recently developed an AI risk framework for tax authorities and tax preparers to help organizations evaluate AI systems before deployment. The conversation examines why extraordinary productivity gains should not automatically mean removing people from the process; how firms can more rigorously evaluate AI vendors; and why accuracy, auditability, transparency, and governance must remain central to AI risk management.

The discussion also considers a significant long-term risk: the erosion of human expertise. As AI agents begin performing work traditionally handled by junior accountants and tax professionals, the profession must determine how the next generation will build the knowledge and judgment required to supervise those systems effectively.

Werfel makes the case that AI should sit beside the tax professional, not take their chair.

Additional topics include:

• AI hallucinations

• human-in-the-loop systems

• client trust

• AI pricing and billing

• and the opportunity for the tax profession to become a model for responsible AI adoption.

For accountants, tax professionals, firm leaders, technology providers, and anyone seeking to understand how AI will reshape the profession, this conversation offers an important counterweight to the race for maximum automation.

Link to Danny’s AI Risk Framework: https://www.thetaxadviser.com/issues/2026/aug/a-risk-framework-for-ai-use-in-tax-administration-and-preparation/

Connect with us here:

• Danny Werfel
• Kacee Johnson⁠⁠
• ⁠Digits
• AI Native Accounting Foundation