Nana Wu
A federal court in the Northern District of California has delivered a significant victory to AI company Anthropic, ruling that directives issued by then-Secretary Pete Hegseth, which designated Anthropic a national security risk and banned its products from federal use, violated the company’s First Amendment rights to free speech and due process, and were an arbitrary and capricious action under the Administrative Procedure Act (APA). This landmark decision, arising from a lawsuit filed by Anthropic challenging the directives, not only vacates the government’s actions but also issues a permanent injunction preventing their re-application.
Background: A Clash Over AI Safeguards and National Security Designations
The dispute originated when Anthropic, the developer of the AI model Claude, refused to comply with demands from Secretary Hegseth and the Department of Defense (referred to by the Trump Administration’s nickname, "Department of War") for a version of its AI that would omit certain safety and ethical safeguards. Anthropic argued that these safeguards were crucial for responsible AI development and deployment, particularly when dealing with sensitive government applications. In response to Anthropic’s principled stance and public articulation of its safety concerns, Hegseth declared the company a supply chain risk to national security. This designation triggered a cascade of restrictive measures: federal agencies were ordered to cease using Anthropic’s products, and defense contractors were prohibited from engaging in any commercial activity with the company, even if unrelated to military contracts.
Anthropic responded by filing a lawsuit in the Northern District of California, asserting that the government’s actions were retaliatory and unconstitutional. This legal challenge culminated in a definitive ruling that largely favored Anthropic, underscoring the critical balance between national security imperatives and the fundamental rights of businesses operating within the United States.
Chronology of the Dispute and Legal Proceedings
The events leading to the court’s decision unfolded over a period, highlighting a rapid escalation of government action against Anthropic.
- Early 2026 (Approximate Timeline): Discussions and negotiations between Anthropic and the Department of Defense regarding the development of a specific version of the Claude AI model for government use. Anthropic expresses concerns and insists on maintaining ethical safeguards.
- February 24, 2026 (Implied): Secretary Hegseth reportedly raises the possibility of invoking the Defense Production Act, a move that, paradoxically, could have suggested Anthropic’s essentiality to national security rather than its being a risk.
- February 27, 2026: Secretary Hegseth issues a directive designating Anthropic as a supply chain risk and ordering all defense contractors to cease commercial activity with the company. This directive is publicly broadcast via social media.
- Following Days/Weeks: Various federal agencies, including Treasury, FHFA, State, GSA, OPM, NRC, DHS, and Energy, issue orders to terminate their use of Anthropic’s products, purportedly in accordance with a presidential directive.
- March 2026: Anthropic files its lawsuit in the Northern District of California, challenging the legality and constitutionality of the government’s actions.
- March 27, 2026 (Previous Victory): The court issues a preliminary injunction against Hegseth and his department, temporarily halting the enforcement of the directives.
- Present Ruling: The court issues a ruling on the merits of Anthropic’s claims, granting a permanent injunction and vacating the challenged agency actions.
Court’s Findings: A Multifaceted Violation of Rights
The court’s decision meticulously dissects the government’s actions, finding them to be in violation of multiple legal and constitutional principles.
First Amendment Violations: Retaliation for Protected Speech
A central tenet of the court’s ruling is that the government’s designation of Anthropic as a national security risk and the subsequent punitive measures constituted unlawful retaliation for the company’s constitutionally protected speech. The court found that Anthropic had engaged in speech on matters of public concern, specifically regarding the safeguards embedded within AI systems and their implications for responsible development. The administration’s own public statements and social media posts, often characterized by strong rhetoric against Anthropic’s "corporate virtue-signaling" and "Silicon Valley ideology," clearly indicated that the designation was intended to punish Anthropic for its public stance and its refusal to compromise on its ethical principles.
The court noted Secretary Hegseth’s explicit linking of Anthropic’s punishment to its public statements, quoting him as saying Anthropic "delivered a master class in arrogance" and attempted to "strong-arm the United States military" through "sanctimonious rhetoric of ‘effective altruism.’" The President’s characterization of Anthropic as a "RADICAL LEFT, WOKE COMPANY" further cemented the court’s finding that the punitive actions were motivated by the government’s displeasure with Anthropic’s viewpoint, rather than by any legitimate national security concern. The court concluded that Anthropic’s protected speech was a substantial motivating factor in the defendants’ actions, and that such actions would not have been taken absent the desire to make an example of the company.
Due Process Violations: Lack of Notice and Opportunity to Be Heard
The court also found that Anthropic was denied its Fifth Amendment due process rights. The ruling emphasized that due process requires notice "reasonably calculated, under all the circumstances, to apprise interested parties of the pendency of [a government] action and afford them an opportunity to present their objections." In this case, Anthropic received neither adequate notice nor a meaningful opportunity to object before being summarily blacklisted and designated a supply chain risk. The company learned of the factual basis for its designation only through the litigation process, a clear violation of its right to pre-deprivation process. The court specifically highlighted that the actions were taken without any meaningful notice or opportunity for Anthropic to present its case, particularly concerning the factual basis for the supply chain risk designation.
Administrative Procedure Act Violations: Arbitrary and Capricious Action
Beyond constitutional infringements, the court determined that the government’s actions were also unlawful under the Administrative Procedure Act (APA). The APA mandates that agency actions must not be "arbitrary, capricious, an abuse of discretion," or "in excess of statutory jurisdiction, authority, or limitations." The court found that Hegseth’s directive and the supply chain designation met these criteria.
Specifically, the court examined 10 U.S.C. § 3252, which defines "supply chain risk" as the risk that an adversary may "sabotage, maliciously introduce unwanted function, or otherwise subvert" a covered system. The court concluded that Anthropic’s public stance on contract terms did not fall within this definition, which is clearly aimed at covert acts, sabotage, or the introduction of malicious code by adversaries, not overt public positions on contractual terms. The legislative history of the statute further supported this interpretation, focusing on risks of counterfeit or malicious code introduced by suppliers.
Furthermore, the court found that Hegseth’s order imposing a secondary boycott on Anthropic exceeded his statutory authority. The order, which effectively banned any contractor doing business with the Department of War from conducting commercial activity with Anthropic, was deemed a final agency action subject to APA review. The court rejected the government’s argument that the order was not reviewable because Hegseth lacked the authority to issue it, stating that such an interpretation would undermine the very purpose of the APA. The court reasoned that if such actions were unreviewable simply because an agency lacked the requisite authority, it would create a loophole for unlawful agency behavior.
Inadequate Statutory Basis for Action
The court also delved into the specifics of the statutes cited by the government. While Anthropic presented arguments regarding an ultra vires claim related to a presidential directive, the court did not rule in its favor on this specific claim, finding that the arguments were not sufficiently developed and that the President’s broader constitutional authority, while subject to limits, was not clearly exceeded in a manner that would render the order entirely void under the presented legal framework. However, this did not detract from Anthropic’s overall victory, as the other claims provided ample grounds for the court to grant relief.
Crucially, the court found that the government had failed to demonstrate any genuine national security interest that outweighed Anthropic’s speech rights. Evidence presented by Anthropic showed inconsistent behavior from the government, including prior discussions about invoking the Defense Production Act (suggesting essentiality) and subsequent discussions about collaboration on new AI models, which contradicted the narrative of Anthropic being a genuine security threat.
Relief Granted: Vacatur and Permanent Injunction
The court’s ruling provided Anthropic with comprehensive relief, including both the vacatur of the government’s prior actions and a permanent injunction against their re-application.
Vacatur of Agency Action
The court disagreed with the government’s request for a remand to the agency for further action, stating that such a remand would be an inadequate remedy. Given Anthropic’s success on constitutional and APA grounds, the court determined that the challenged actions were fundamentally flawed and could not be cured by further administrative deliberation. The court noted that there was no evidence that vacatur would result in a national security risk, especially since the directives had already been preliminarily enjoined for over five months without apparent harm.
Permanent Injunction
In addition to vacating the prior directives, the court granted a permanent injunction. This was deemed necessary because vacatur alone would not fully redress the constitutional injury Anthropic had suffered. The court acknowledged that the government might argue that its challenged conduct had ceased voluntarily. However, citing precedent, the court found that it was not "absolutely clear that the allegedly wrongful behavior could not reasonably be expected to recur." The government’s continued defense of its actions and refusal to stipulate against future similar conduct indicated a potential for recurrence, thus necessitating injunctive relief to ensure Anthropic’s rights were protected going forward. The court specifically noted that the argument that Anthropic "continues to speak freely" and has seen its valuation increase was a direct result of the preliminary injunction restoring the status quo, and that lifting that injunction could expose Anthropic to renewed harm.
Implications and Broader Impact
This ruling has significant implications for the relationship between government agencies, technology companies, and the First Amendment, particularly in the rapidly evolving field of artificial intelligence.
Protecting Free Speech in Government Contracting
The decision firmly establishes that government agencies cannot retaliate against companies for their public statements or for refusing to compromise on ethical principles, even within the context of government contracting. The court’s recognition that discussing AI safeguards constitutes speech on a matter of public concern is particularly important as AI technology becomes more integrated into critical infrastructure and defense systems.
Limits on National Security Designations
The ruling sets a precedent for challenging government designations of companies as national security risks, particularly when such designations appear to be retaliatory rather than based on genuine, demonstrable threats. The court’s scrutiny of the government’s actions and its finding that the asserted national security concerns were contradicted by subsequent government behavior signal a higher bar for agencies seeking to use such designations to stifle dissent or enforce policy preferences.
Due Process in Agency Actions
The emphasis on due process rights serves as a crucial reminder that even in matters of national security, agencies must adhere to fundamental fairness. The requirement for notice and an opportunity to be heard before imposing severe sanctions like blacklisting is a cornerstone of administrative law.
The Future of AI Development and Regulation
This case highlights the ongoing tension between rapid technological advancement and the need for robust ethical frameworks and regulatory oversight. Anthropic’s victory suggests that companies committed to responsible AI development can find legal recourse against government pressure to compromise on safety and ethics. It underscores the importance of clear statutory definitions for terms like "supply chain risk" and the need for agencies to act within their delegated authorities.
The court’s order explicitly states that this ruling does not prevent the Department of War from taking lawful actions consistent with applicable regulations, statutes, and constitutional provisions. This means the department remains free to choose its AI vendors and transition to other providers, provided these actions are not retaliatory or otherwise unlawful. The case serves as a powerful reminder that even in sensitive areas like national security and advanced technology, the Constitution remains the ultimate arbiter of government power.
Joko Widodo
admin
Lina Irawan
Yuyun Lusiyani
Pevita Pearce
Eva Lovia