Identifying “item enemies”—items that cue or overlap with one another—is one of the most time- and labor-intensive tasks facing subject matter experts (SMEs) during test assembly. This session presents a pilot study examining whether artificial intelligence can reliably identify item enemies, offering the potential for significant time savings without sacrificing content quality. Using an item bank for a mid-size certification organization in the medical field, a generative AI tool reviewed 448 items for cueing and overlap, assigning each pairwise comparison a confidence level and a written rationale. Three SMEs will independently review a sample of item pairs, comparing their judgments to the tool’s classifications before and after seeing its rationale for classifications. Attendees will learn how the AI classification tool was developed, how confidence levels were determined, and how the SME panel judged whether AI is ready to support item enemy identification going forward.