An AI startup has sued the U.S. government after losing access to Anthropic’s most advanced AI models.
Legal tech company Legion filed a lawsuit in Washington, D.C., arguing that a government directive forcing Anthropic to restrict access to its Fable 5 and Mythos 5 models caused severe harm to its business.
Legion said its AI-powered legal software relies on Anthropic’s frontier models and that the restrictions instantly cut off a critical tool used to develop and operate its platform. The company described the impact as “immediate, irreparable, and existential” due to the rapid pace of AI competition.
The case stems from a recent government order requiring Anthropic to block foreign nationals from accessing the models over national security concerns. Anthropic initially suspended access entirely before later restoring limited access with nationality-based controls and additional compliance checks.
Legion, which employs Canadian workers remotely from Canada, argues that the restrictions unfairly disrupted its operations despite being a U.S.-based company.
The lawsuit adds a new dimension to Anthropic’s ongoing dispute with the Trump administration, with critics arguing the restrictions were excessive while supporters say powerful AI systems deserve heightened scrutiny. Some cybersecurity experts have also urged the government to reverse the order, warning it could weaken defensive cybersecurity capabilities in the United States.