<p>This paper explores the tensions that arise in applying our received notions of artistic creation and credit to AI-generated art. Focusing on AI-generated images, I articulate an “antinomy of artificial art”—a contradiction between three plausible propositions: (1) AI-aided images can count as works of art, (2) All artworks have artists, and (3) AI-aided images don’t (clearly) have artists. To support (3), I argue that, in the case of many AI-aided works, none of the potential candidates—the AI system itself, its programmers, the creators of training images, or the human prompter—can be unproblematically credited as artist. I explore three intriguing responses to this antinomy: treating the prompter as an institutional artist, crediting the AI as a creative agent, and viewing AI–human collaboration as an artistic collective. While each approach shows promise, I argue that none succeeds. If I’m right, current generative AIs inhabit a sort of conceptual limbo. They are “super-tools” possessing sophisticated generative capacities but lacking the robust intentionality and normative responsiveness required for prototypical artistic creation. The tension they expose in our conceptual scheme can’t be resolved through further empirical or theoretical investigation, I suggest. Rather it will be resolved through a socially-driven revision of our artistic concepts and institutions as we adjust them to our developing AI-aided artistic practices.</p>