Generative processes have captured the public attention in recent years, from the fever dream images of early neural nets to the more recent proliferation of chatbots and language models. In the world of cataloging, a 2021 post on the PCCLIST cataloging listserv about a book "co-authored" by a transformer language model led to an almost week-long discussion over whether the AI was truly an author or just a tool. But generative texts are not new, and catalogers are no strangers to determining who counts as an author: the question of whether non-humans, including animals and fictional characters, can author a work of literature has been a topic of intense deliberation as the cataloging world moves toward the implementation of the Official RDA Toolkit, a cataloging standard based on the Library Reference Model (LRM). The LRM holds that fictional characters, and non-human entities more broadly, cannot be Persons or Agents (and for good measure, that fictional places cannot be Places); however, catalogers are still faced with books written by mouse detectives, starship captains, ghosts, Muppets, and presidential pets. How does a cataloger balance faithfully describing an item as it represents itself with following the rules? Who counts as an author? This presentation will examine the ways that non-human creators are credited in catalog records, looking at the connections between fictional and animal authorship, automatic writing, computer-generated texts, and more. What is a tool, what is a process, and what is an author - at least, according to modern cataloging standards?