As AI becomes a routine part of academic and professional writing, assessment programs face a pressing question: should we focus on detecting AI, or on clarifying what we actually want to measure? This panel reframes writing assessment for the age of AI co-authorship by shifting attention from authorship detection to construct definition. Panelists will explore how assessments can distinguish among independent writing, revision skill, source use, argumentation, prompt engineering, ethical AI use, and effective communication with tools. Participants will leave with practical language and design considerations for evaluating whether current writing assessments still measure what they claim to measure—and how future assessments can better reflect the realities of writing today.