AI Model Gatekeeping: Anthropic and US Government Demonstrate Control Over Frontier Access

Over the past two weeks, Anthropic and the U.S. Government took actions demonstrating their ability to restrict access to frontier AI models. Anthropic released Fable 5 with restrictions preventing others from using the model to develop competing LLM technology, citing safety concerns. Shortly after, the U.S. Government imposed export controls on Fable and Mythos, requiring special licenses for foreign nationals to access them. Anthropic subsequently disabled global access to Fable. These developments created uncertainty among developers regarding the stability of building on proprietary AI platforms. Sudden policy changes and undisclosed restrictions raised concerns about the reliability of relying on single-model providers. The actions also prompted discussions about AI sovereignty among nations worldwide. Countries recognized that American companies and government can quickly limit global access to critical technology, spurring investments in open-source alternatives and independent model development, despite the considerable challenges in training frontier models.