Amid expanding government investments in artificial intelligence (AI), questions about sovereign AI have become more pressing than ever. What exactly is sovereign AI? To what degree can sovereignty be achieved? And what does it take to build it? These questions have gained urgency following events such as the June 2026 case in which Anthropic was forced to suspend access to its latest AI model. The suspension followed a U.S. export control directive barring use by non-U.S. persons, including foreign nationals within its own U.S. offices. Cases like this highlight how a government’s policies can directly affect the global use of frontier AI systems.