Anthropic’s complaint alleges that the Trump Administration and multiple federal agencies unlawfully retaliated against the company for refusing to let its Claude AI models be used for fully autonomous lethal weapons and mass surveillance of Americans. According to the filing, Anthropic had become a key AI provider to the Department of War, including on classified systems, and had worked with the government under a detailed usage policy that barred those two categories of use while otherwise tailoring “Claude Gov” models to national security needs.
The dispute escalated when the Department of War demanded that Anthropic drop its remaining two restrictions and allow “all lawful use” of Claude, including for lethal autonomous warfare and large-scale domestic surveillance. Anthropic’s CEO publicly stated the company could not “in good conscience” agree, citing safety, reliability, and civil-liberties concerns given Claude’s limitations and the risks of agentic, high-speed data analysis. The complaint describes public criticism from senior defense officials, who allegedly called the CEO ideological and unpatriotic and issued an ultimatum that Anthropic either “get on board” or face exclusion from the defense supply chain or compelled production under the Defense Production Act.
After Anthropic held its position, President Trump allegedly posted a directive on social media ordering every federal agency to “IMMEDIATELY CEASE all use of Anthropic’s technology,” denouncing the company as a “radical left, woke” actor and threatening to use the “full power of the Presidency” and potential civil and criminal consequences. The Secretary of War then announced via social media that Anthropic was being designated a “Supply-Chain Risk to National Security,” barred all military contractors and partners from doing any commercial business with Anthropic, and later sent a formal letter invoking 10 U.S.C. § 3252 and asserting that no less intrusive mitigation measures were available. Other agencies—including GSA, Treasury, and the Federal Housing Finance Agency—are alleged to have followed by canceling contracts or directing staff to stop using Anthropic’s services, cutting off major federal business and allegedly chilling private-sector relationships as well.
Legally, Anthropic argues these “Challenged Actions” violate multiple doctrines: that the Secretary’s supply-chain designation and related bans exceed and misapply 10 U.S.C. § 3252 and are arbitrary and procedurally defective under the Administrative Procedure Act; that the President and agencies acted ultra vires because no statute authorizes a blanket government-wide cutoff solely due to a vendor’s policy-based speech; that the measures constitute unlawful First Amendment retaliation for Anthropic’s public and private advocacy about AI safety and its own product’s limitations; that they deprive Anthropic of property and liberty interests—reputation, ongoing and future business, and its ability to advocate—without due process under the Fifth Amendment; and that they amount to sanctions and penalties imposed without lawful authority in violation of the APA’s limits on agency power. The company claims immediate and irreparable harm to its business, reputation, and speech rights, and asks the court to declare the President’s and agencies’ actions unlawful and enjoin any further efforts to implement or enforce them.
xvxvx•2h ago
Anthropic’s complaint alleges that the Trump Administration and multiple federal agencies unlawfully retaliated against the company for refusing to let its Claude AI models be used for fully autonomous lethal weapons and mass surveillance of Americans. According to the filing, Anthropic had become a key AI provider to the Department of War, including on classified systems, and had worked with the government under a detailed usage policy that barred those two categories of use while otherwise tailoring “Claude Gov” models to national security needs.
The dispute escalated when the Department of War demanded that Anthropic drop its remaining two restrictions and allow “all lawful use” of Claude, including for lethal autonomous warfare and large-scale domestic surveillance. Anthropic’s CEO publicly stated the company could not “in good conscience” agree, citing safety, reliability, and civil-liberties concerns given Claude’s limitations and the risks of agentic, high-speed data analysis. The complaint describes public criticism from senior defense officials, who allegedly called the CEO ideological and unpatriotic and issued an ultimatum that Anthropic either “get on board” or face exclusion from the defense supply chain or compelled production under the Defense Production Act.
After Anthropic held its position, President Trump allegedly posted a directive on social media ordering every federal agency to “IMMEDIATELY CEASE all use of Anthropic’s technology,” denouncing the company as a “radical left, woke” actor and threatening to use the “full power of the Presidency” and potential civil and criminal consequences. The Secretary of War then announced via social media that Anthropic was being designated a “Supply-Chain Risk to National Security,” barred all military contractors and partners from doing any commercial business with Anthropic, and later sent a formal letter invoking 10 U.S.C. § 3252 and asserting that no less intrusive mitigation measures were available. Other agencies—including GSA, Treasury, and the Federal Housing Finance Agency—are alleged to have followed by canceling contracts or directing staff to stop using Anthropic’s services, cutting off major federal business and allegedly chilling private-sector relationships as well.
Legally, Anthropic argues these “Challenged Actions” violate multiple doctrines: that the Secretary’s supply-chain designation and related bans exceed and misapply 10 U.S.C. § 3252 and are arbitrary and procedurally defective under the Administrative Procedure Act; that the President and agencies acted ultra vires because no statute authorizes a blanket government-wide cutoff solely due to a vendor’s policy-based speech; that the measures constitute unlawful First Amendment retaliation for Anthropic’s public and private advocacy about AI safety and its own product’s limitations; that they deprive Anthropic of property and liberty interests—reputation, ongoing and future business, and its ability to advocate—without due process under the Fifth Amendment; and that they amount to sanctions and penalties imposed without lawful authority in violation of the APA’s limits on agency power. The company claims immediate and irreparable harm to its business, reputation, and speech rights, and asks the court to declare the President’s and agencies’ actions unlawful and enjoin any further efforts to implement or enforce them.