<p>When it comes to global coordination and global governance of AI, it is reasonable to suppose that we need a global ethics to guide it. But the idea of a global ethics of AI raises several philosophical and practical challenges which, despite the existence of AI ethics guidelines and related governance documents with a supranational character such as those from the European Commission, the OECD, UNESCO, and the UN, have received insufficient discussion. This paper argues that a truly <i>global</i> ethics of AI that aims to be effective and morally adequate must overcome at least three fundamental philosophical and practical, political challenges related to cultural differences, geopolitics and ideologies, and anthropocentrism. The paper offers some conceptual resources that may help to navigate these difficulties. The analysis and discussion provided in this paper can be used to critically assess existing AI ethics guidelines and offers elements of a vision for a global AI ethics that is more inclusive, more open, and more radically relational.</p>