The Trump administration ordered Anthropic to cut access to its AI models for all foreign nationals, including those inside the US and its own employees. Experts criticize this as an unprecedented, opaque, and ad hoc approach to AI governance.
The Trump administration issued an export control order to Anthropic, forcing the company to block access to its latest AI models, Fable 5 and Mythos 5, for all foreign nationals, including those residing in the US and the company's own foreign employees. Anthropic complied by cutting off all access to the models.
The administration cited 'national security authorities' for the export control directive but did not publicly explain the legal basis. Anthropic stated that the government was concerned about a potential 'jailbreak' by China-linked groups, but Anthropic argued that the jailbreak did not bypass all safeguards. This is the first time US export controls have been used to restrict access to an AI model, marking an unprecedented move.
The episode highlights an opaque and ad hoc approach to AI governance, which experts warn is unsustainable. It underscores the need for a clear legal framework and consistent policy. Additionally, the ambiguity of what constitutes 'export' in the context of AI models raises further questions.