This article seeks to clarify the relation between AI and intellectual property (IP) in the information society. It aims to critically examine our IP framework at the dawn of the Fourth Industrial Revolution. In that context, it contends that human authorship and inventorship remain the normative wellsprings of IP law. Additionally, it argues that extending copyrights to AI works hinders innovation, cultural diversity and even fundamental freedoms. Adding extra layers to the existing rainbow of IP rights is not a good solution to balance the societal impact of technological progress. Legislative gaps can be remedied with contracts and generous application of fair use and the three-step-test in international IP law. Finally, parts of the Roman multi-layered property architecture can be relevant for AI. Building upon this framework, the article includes a proposal for a new public domain model for AI creations and inventions that crossed the autonomy threshold: Res Publicae ex Machina (Public Property from the Machine). Introducing the legal concept of Public Property from the Machine revitalizes the public domain and its paradigms of liberty, equality, and openness. It is this safe, permission-free space where creativity and inventiveness can flourish, benefitting open innovation, culture, and society at large.