Michael V. Sala, AI Won’t Shrink Corporate Tax Teams Much. Here’s Why., 121 Tax Notes State 189 (July 20, 2026)

In this article, Sala argues that the anticipated effects of artificial intelligence agents in modern corporate tax functions are greatly overstated because AI ultimately cannot shoulder regulatory, ethical, or financial consequences of culpability.

As applied to the modern corporate tax function, these models mistakenly conflate tax preparation with tax practice. Tax preparation involves the routine movement of data into forms. Tax practice, on the other hand, involves facilitating engagement between a business, the tax authorities it’s subject to, its financial statement auditors, and its executive leadership team. Practitioners in modern corporate tax functions understand that manual form preparation has become rare in all but a few cases. Tax return automation systems are already both widely available and in place, and extensive statutory and legal interpretive analysis was and will continue to be performed in the national practices of the Big Four. The modern corporate tax function hasn’t been a tax return preparation or memo writing pool for decades, so the notion that it is about to convert from that role through a step-change in technology is misguided.

In fact, many of the skills and tasks that AI is best at are now simply redundant additions to an already teched-out corporate tax function. 

The demonstration also implicated a much more expansive use of the model in complex tax planning domains that are managed primarily in the national offices of Big Four tax and law firms. Could tax AI agents replace or dislodge the tendency for U.S. corporates to wait for or defer to an expensive Big Four opinion on a tax planning matter by instead using a highly technical tax AI for direction? Or are the tasks that the Big Four complete about more than the delivery of an accurate opinion on the law?