The paper explores the evolving concepts of authorship and creativity in AI-generated literary texts. By examining AI generation methods and the theoretical frameworks surrounding AI literature, it demonstrates how AI-generated works challenge traditional notions of literary authorship. The natural language and narrative or poetic structures used by AI provoke empathetic responses from readers, fostering a deceptive relationship and breaking the hermeneutic contract. However, human input remains crucial, prompting a reconceptualization of authorship as a collaborative process, a perspective first proposed by Italo Calvino and later revisited by contemporary AI novelists. The second part of the paper focuses on the notion of hallucination, which refers to errors in meaning-making by AI systems that can serve as expressive opportunities for artistic creation. These hallucinations link text generation to human creativity. First explored by visual artists, this phenomenon has recently gained attention in literary contexts, notably in the works of Ross Goodwin and Rocco Tanica. Hallucination as a creative principle parallels literary development driven by deviation from norms. The paper illustrates this potential with examples of AI-generated texts.