In a piece for Texas Lawyer, Steno Director of Legal Solutions Joe Stephens, JD, examines a new law review essay by Andrea Roth and Rebecca Wexler that proposes a fresh legal framework for AI privilege claims.

The essay's authors suggest applying the long-established Kovel necessity test, which extends privilege to non-lawyers when their presence is necessary for the lawyer to do the job. However, Joe, who runs a rural West Texas public defender office, argues that the test creates an uncomfortable structural gradient. Proving necessity requires an attorney to file a sworn affidavit documenting their own office's shortcomings.

Here's how Joe describes what that filing could actually say:

Counsel currently carries 280 open felony cases. Counsel was unable to meet with the defendant for more than twenty minutes before his preliminary hearing, and lacked the staff resources to review the discovery with him personally, making the client's independent review of the police reports through an AI tool necessary to the representation.

That same affidavit, Joe notes, could later surface in a client's ineffective-assistance appeal. The dynamic plays out differently depending on who the client is. A defendant with a full legal team on call will likely lose a privilege claim over their AI chat logs because a court can reasonably conclude they didn't need the chatbot when they had lawyers available. An indigent defendant represented by an overloaded public defender is in the opposite position.

The bigger the caseload, the stronger the case that AI use was genuinely necessary. As Joe puts it, "the degree of privilege protection increases in direct proportion to systemic failure." He then closes with three practical steps that he argues public defenders need to take now, covering client intake, data-retention settings, and real-time documentation of staffing shortfalls.

Read the full piece on Texas Lawyer.