“Critical AI Studies has consolidated into a distinct humanities field, and its central claims bear directly on questions art history has long made its own: that AI is a material and extractive infrastructure, that training data constitutes a kind of quasi-archive, that “latent space” is a new aesthetic and epistemic condition, and that generative images inaugurate a crisis in the photographic index. The panel treats AI as an epistemic regime whose aesthetic, economic, and infrastructural logics demand critical analysis, refusing the dominant binary of AI as existential threat versus AI as neutral tool in favor of questions grounded in art-historical debate. Its lines of inquiry include what an image is once it is synthesized from a latent distribution rather than made or taken; how latent space might be analyzed as an aesthetic and epistemic condition; how art-historical theories of the archive can sharpen our understanding of the training dataset as both canon and quasi-archive; whether “AI slop” and model collapse should be historicized as a style, and within what genealogies; how synthetic media reorganize photographic indexicality; what an art history of artist refusal and counter-datasets might look like; how AI’s material substrate of data labor, compute, and mineral extraction enters art-historical analysis; and what sovereign datasets propose for inherited questions of canon and authorship.” (Session description)