Anthropic recently took its AI models, Claude Fable 5 and Mythos 5, offline amid U.S. export control warnings. The company had warned that these powerful tools could be misused by malicious actors, highlighting the dual-use nature of advanced AI.

The government’s move came after concerns were raised about disabling the guardrails on Fable 5 to unlock full Mythos 5 capabilities, seen as a potential national security risk.

Despite ongoing talks with the White House, Anthropic remains under pressure. Meanwhile, experts predict that similar capabilities will become widespread among AI models from various companies in the near future.

This institutional clash might merely delay or mask the inevitable: as AI advances, so too do its risks and rewards. The race to develop and control these tools is on, with no clear winner yet.