Foley defines a four-part test for defensible AI workflows

On October 1, 2026, Foley & Lardner published an operating framework built around design, authorization, proof and accountability for AI-enabled legal work. The firm says a defensible workflow should record source materials, model outputs, attorney revisions and supporting authority, and should identify where responsibility passes between the system and the lawyer. That approach turns AI strategy from tool access into an auditable production method that can be reproduced across matters. Firms adopting similar controls will be better positioned to answer client questionnaires, defend work quality and measure whether a workflow is actually improving delivery.