Legion v. United States (Anthropic export-directive challenge)

Legion v. United States (No. 1:26-cv-02225, D.D.C., filed June 23, 2026) is a federal suit brought by Legion LegalTech Corp., a San Jose legal-technology company, challenging a June 12, 2026 Commerce Department directive that required Anthropic to disable its Fable 5 and Mythos 5 models for "any foreign national," a restriction that in practice took both models offline worldwide. The complaint asks the court to vacate the directive and signals that Legion will seek a preliminary injunction barring its enforcement. It is the first reported court challenge to the federal government's use of export-control authority to withdraw a commercial AI model — rather than hardware — from the market (Source: thenextweb.com). Anthropic is not a party to the suit (Source: thenextweb.com).

Infobox

Background

On June 12, 2026, the Bureau of Industry and Security at the U.S. Department of Commerce issued a directive citing national security that required Anthropic to disable Fable 5 and Mythos 5 for any foreign national, inside or outside the United States, including Anthropic's own foreign-national staff. Because the company could not verify nationality in real time, the only way to comply was to switch both models off for all users the same day; Anthropic characterized the action as disproportionate (Source: anthropic.com). According to Anthropic's account, the government's concern arose from a technique that involved asking the model to read a codebase and identify flaws — a capability Anthropic described as widely available in other models and used routinely by cyber defenders (Source: anthropic.com).

Legion builds drafting and case-management tools for litigators and depends on Anthropic's models for its platform. Its software-development team includes Canadian nationals working from Canada — the category the directive targets — so the shutdown removed the team's access to the models the product is built on. Legion's complaint describes the resulting harm as "immediate, irreparable, and existential," arguing that "competitive ground lost during a suspension cannot be regained after the fact" (Source: thenextweb.com).

Claims and relief sought

Legion asks the court to set aside the June 12 directive and says it will seek a preliminary order freezing enforcement while the case proceeds (Source: thenextweb.com). The complaint (Complaint, Legion LegalTech Corp. v. United States (D.D.C., June 23, 2026)) pleads three causes of action: that the directive exceeds export-control authority under ECRA, because "the only export-control classification that ever directly covered advanced AI model weights — ECCN 4E091 — was rescinded in May 2025 with no replacement" and because providing access to a hosted model "releases no technology, source code, model weights, or technical data to the user" and so falls outside the deemed-export framework; that to the extent it rests on IEEPA it violates the Berman Amendment, 50 U.S.C. § 1702(b)(3), which "categorically excludes" any power to regulate the export of informational materials, into which category the complaint places the model's "drafted text, written legal analysis, summaries, and similar composed material"; and that it is arbitrary and capricious. The underinclusiveness allegation is that "the identical code-analysis capability remains publicly available through competing products the directive left untouched, including OpenAI's GPT-5.5," and that less-restrictive alternatives such as restricting access only for nationals of adversary nations went unconsidered. On information and belief, the complaint alleges the directive "was precipitated not by any neutral agency assessment but by telephone calls from Anthropic's commercial rivals over a single evening, followed by a ninety-minute ultimatum and no disclosed independent analysis" (Complaint, Legion LegalTech Corp. v. United States (D.D.C., June 23, 2026)).

The plaintiff's public account frames the directive as procedurally and constitutionally defective: it points to a June 2, 2026 executive order, "Promoting Advanced Artificial Intelligence Innovation and Security," which expressly provided that nothing in it authorizes a mandatory licensing, preclearance, or permitting requirement for releasing AI models, and argues that the government imposed exactly such a requirement on a single company ten days later (Source: legion.law). The account also invokes an earlier judicial finding — from the separate Anthropic v. U.S. government matter, in which a court found the administration's stated national-security reasons for an earlier action against Anthropic were "pretextual" and the "real motive was unlawful retaliation" — to characterize the directive as discretion exercised without limit or accountability (Source: legion.law).

Provenance note: The complaint text cited here was retrieved from the plaintiff's own domain; the as-filed version should be confirmed against the D.D.C. docket.

Legion continued pressing the suit after the June 30, 2026 withdrawal of the export controls, arguing in accounts detailed July 3, 2026 that the administration exceeded its legal authority in imposing the controls and emphasizing that its API use of the model involved no access to model weights (Source: insideaipolicy.com). Researchers separately floated First Amendment challenges to the controls in reporting from June 29, 2026, framing the withdrawal of model access as a restraint on protected expression (Source: politico.com).

Significance and related proceedings

The episode is the first time the federal government has used export-control authority to pull a commercial AI product, rather than hardware, off the market (Source: thenextweb.com). The suit joins an existing posture of conflict between Anthropic and the administration: the two are already in litigation in Washington and California after Anthropic sued over the Pentagon's designation of it as a national-security supply-chain risk, which followed Anthropic's refusal to permit military use of Claude for autonomous weapons or domestic surveillance (Anthropic v. United States (Pentagon ban challenge)) (Source: thenextweb.com). The directive's reach extended to allied users, including the United Kingdom's AI Security Institute, which had been evaluating the affected models when access was cut off (Source: thenextweb.com). Reporting noted that the administration restricting the models was the same one that had urged banks to adopt AI and had earlier opposed expanding civilian access to Mythos even as the National Security Agency continued using it (Source: thenextweb.com). As of the filing, the government had not responded and both models remained offline (Source: msn.com).

Relationships

• litigates: Anthropic (model developer; not a party)
• instance-of: Export Controls (AI)
• related: Anthropic v. United States (Pentagon ban challenge), Claude Fable 5, Claude Mythos Preview, UK AI Safety Institute (AI Security Institute), Anthropic