Chris Wood
A U.S. legal technology company has sued the federal government over an order that forced Anthropic to shut off access to two of its most powerful AI models for users worldwide, raising urgent questions about government authority over AI export controls and the companies caught in the crossfire.
At a Glance
- Legion LegalTech Corp filed suit in Washington, D.C., federal court challenging a June 12 Commerce Department directive.
- The order required Anthropic to disable its Fable 5 and Mythos 5 models for any foreign national.
- Anthropic cut off all customers the same day to ensure compliance.
- Legion says the shutdown immediately severed access for its Canada-based development team and threatens its core business.
- Anthropic is not a party to Legion's lawsuit but is separately suing the Trump administration in federal court.
What the Commerce Department Order Actually Did
On June 12, the Bureau of Industry and Security, a division of the U.S. Commerce Department, issued a directive requiring Anthropic to block its Fable 5 and Mythos 5 models from "any foreign national." Anthropic responded the same day by cutting off access for all of its customers globally, apparently choosing blanket compliance over a more targeted restriction. The Commerce Department and the White House did not immediately respond to requests for comment.
The legal authority the government relied on is not spelled out in the lawsuit, but the Bureau of Industry and Security typically operates under the Export Administration Regulations, a body of federal rules that restricts the transfer of controlled technology to foreign persons or entities. Export controls on AI models are a relatively new and contested frontier, and courts have yet to clearly define how far that authority extends over software services delivered over the internet.
It is worth keeping in mind that rules governing AI exports, access restrictions, and national security reviews vary significantly depending on jurisdiction, the nature of the technology, and the specific legal authority invoked. What applies here under federal U.S. law would not necessarily govern similar disputes in other countries.
Why Legion Says the Harm Is Existential
San Jose, California-based Legion LegalTech builds drafting and case management tools for attorneys. The company says it relies on Anthropic's platform as a core component of its software. When Anthropic pulled the plug, Legion's Canada-based software development team lost access immediately.
The lawsuit is blunt about the stakes. "The harm to Legion is immediate, irreparable, and existential," the complaint states. "The pace of frontier AI advancement is blistering, and competitive ground lost during a suspension cannot be regained after the fact." That framing matters legally: to win emergency relief, a plaintiff generally must show irreparable harm, a likelihood of success on the merits, that the balance of hardships favors relief, and that the public interest would not be harmed. Courts apply these four factors under the standard set out in cases interpreting Federal Rule of Civil Procedure 65.
Legion asked the court to vacate and set aside the administration's directive and said it would seek a preliminary injunction barring enforcement while the case proceeds.
Anthropic's Separate Legal Fight
Anthropic itself is already in court against the Trump administration, with parallel cases running in both Washington and California federal courts. The company sued after the government moved to place it on a supply chain blacklist. The trigger, according to prior reporting, was Anthropic's refusal to allow the military to use its AI models for domestic surveillance or fully autonomous weapons systems.
Anthropic is not a defendant or plaintiff in Legion's case. On Tuesday, the company pointed to an earlier statement saying it was "grateful to the administration for their ongoing partnership in working to get this matter resolved as quickly as possible," a notably measured tone for a company simultaneously suing the same government.
Frequently Asked Questions
Can the U.S. government legally restrict access to AI models?
Federal law gives the Commerce Department broad authority under the Export Administration Regulations to control the transfer of certain technologies to foreign nationals. Whether that authority extends to AI software services provided over the internet is a live and unsettled legal question that courts are now beginning to address.
What is Legion asking the court to do?
Legion is seeking two things: a permanent order vacating the directive that prompted Anthropic's shutdown, and a preliminary injunction that would block the government from enforcing the directive while the lawsuit is pending.
Does this affect all Anthropic customers?
As of June 12, Anthropic disabled access to Fable 5 and Mythos 5 for all customers, not just foreign nationals, apparently to ensure full compliance with the order. Whether and when access is restored depends on the outcome of ongoing legal and administrative proceedings.
Do export control rules apply differently in other countries?
Yes. U.S. export control law applies to U.S. companies and, in some cases, to foreign entities dealing in U.S. technology, but other jurisdictions have their own frameworks. The rules governing AI access, data transfers, and national security reviews differ considerably from country to country.
What Comes Next
Legion's request for emergency relief will likely be the first test of whether a court is willing to intervene while the broader legal questions get sorted out. The parallel Anthropic litigation adds pressure from another direction. Together, the cases signal that federal AI export controls are moving from a policy debate into active litigation, with real companies arguing that regulatory overreach is causing concrete, immediate harm.
This article is general information only and does not constitute legal advice. Laws and regulations vary by jurisdiction. Consult a qualified attorney for guidance specific to your situation.
Connor Simmons
Frank Carr
Sarah Ross