Anthropic loses legal fight to shed DOD's designation as a 'supply-chain risk'

CFOTO/Future Publishing via Getty Images

The label is more typically assigned to companies suspected of ties to adversarial governments.

Anthropic will remain a Pentagon-designated supply-chain risk, thanks to federal judges who ruled that the Defense Department can prohibit its employees and contractors from using the AI company's products for DOD business..

The designation has been in force since March, when the department sought to alter its contract with Anthropic to replace prohibitions on using its tools for domestic mass surveillance and fully autonomous weapons with allowances for “all lawful use."

The department has used two statutory justifications for labeling Anthropic a supply-chain risk, a designation typically applied to firms suspected to be working for adversarial governments. Anthropic beat back one of the justifications in an August ruling in the Northern District of California, but did not prevail in a lawsuit heard by the D.C. Circuit Court of Appeals.

Friday's ruling was issued by Judges Gregory Katsas and Neomi Rao, both appointed during the first Trump administration, with a dissent by Judge Karen LeCraft Henderson, who was appointed by the first President Bush.

“This was not an unexpected outcome given the composition of the panel that heard the case” said Charlie Bullock, senior research fellow at the Institute for Law & AI think tank. “Historically, both Katsas and Rao have shown a great deal of deference to the Trump administration's assertions of executive authority on matters of national security. If Anthropic had drawn almost any other two judges from the D.C. Circuit for that panel they would have stood a good chance of winning, because I think their legal position is overwhelmingly strong.”

“I think Anthropic’s chances on appeal if they get to appeal either to the en banc D.C. Circuit or the Supreme Court are good. But they’re not guaranteed to be heard, because the appeals process is discretionary,” Bullock said.

In a statement to The Washington Sun , an Anthropic spokesperson said, “We respectfully disagree with the court’s decision. Another federal court has already held the government’s parallel designation unlawful.”

Anthropic added, “We remain confident in our position and are considering all options, including further review.”

Defense Undersecretary Emil Michael wrote in a statement on X, “The hammer of justice has smashed @AnthropicAI arguments. They are a supply-chain risk to the defense industrial base serving the @DeptofWar. Warfighters will sleep better knowing that no private company will insert their opinions in the chain of command. @SecWar was right!”

The ruling will likely keep the Pentagon and its other contractors reducing their use of Anthropic's products, such as its Claude tools. Overall, however, the company's revenue is growing about tenfold per year, and plans an IPO in November that some expect to value the company at a record $2 trillion .

Michael and Defense Secretary Pete Hegseth have been critical of Anthropic in public statements. In January, Hegseth stated in remarks at a SpaceX event, “Department of War AI will not be woke. It will work for us. We're building war-ready weapons and systems, not chatbots for an Ivy League faculty lounge.”

DOD issued an ultimatum on Feb. 24 demanding the company permit its AI to be used for all legal purposes by Feb. 27, which Anthropic refused. Hegseth announced his direction that DOD designate Anthropic a supply-chain risk Feb. 27, and it was formally announced on March 3.

NEXT STORY: Stolen FBI data reveals employees’ roles in intelligence and surveillance

CFOTO/Future Publishing via Getty Images

该标签通常用于怀疑与敌对政府有联系的公司。

由于联邦法官裁定国防部可以禁止其雇员和承包商将人工智能公司 Anthropic 的产品用于国防部业务，因此 Anthropic 仍将被五角大楼列为供应链风险。

该指定自 3 月份起生效，当时该部门试图修改与 Anthropic 公司的合同，以允许“所有合法用途”取代禁止将其工具用于国内大规模监控和完全自主武器的条款。

该部门援引两项法律依据，将Anthropic公司列为供应链风险企业，这一认定通常适用于被怀疑为敌对政府效力的公司。Anthropic公司在加州北区地方法院8月份的一项裁决中驳斥了其中一项法律依据，但在哥伦比亚特区巡回上诉法院审理的诉讼中败诉。

周五的裁决由格雷戈里·卡特萨斯法官和内奥米·拉奥法官作出，这两位法官都是在特朗普第一任总统执政期间任命的；而卡伦·勒克拉夫特·亨德森法官则持反对意见，她是由老布什总统任命的。

“考虑到审理此案的合议庭组成，这个结果并不出人意料，”法律与人工智能研究所高级研究员查理·布洛克表示。“从历史来看，卡特萨斯和拉奥两位法官都对特朗普政府在国家安全问题上行使的行政权力表现出极大的尊重。如果安斯罗派克公司从哥伦比亚特区巡回上诉法院挑选其他两位法官组成合议庭，他们很有可能胜诉，因为我认为他们的法律立场非常强硬。”

“我认为，如果Anthropic公司能够上诉到哥伦比亚特区巡回上诉法院全体法官席位或者最高法院，他们的上诉成功几率很大。但他们的上诉不一定会被受理，因为上诉程序具有自由裁量权，”布洛克说。

人类学协会的一位发言人在给《华盛顿太阳报》的一份声明中表示：“我们不同意法院的裁决。另一家联邦法院已经裁定政府的平行指定是非法的。”

Anthropic补充道：“我们仍然对自己的立场充满信心，并正在考虑所有选项，包括进一步审查。”

国防部副部长埃米尔·迈克尔在X网站上发表声明称：“正义之锤粉碎了@AnthropicAI的论点。他们对服务于@DeptofWar的国防工业基础构成供应链风险。作战人员可以安心入睡了，因为没有私营公司能够将他们的意见强加于指挥系统之上。@SecWar是对的！”

这项裁决很可能会促使五角大楼及其其他承包商减少对安特罗皮克公司产品（例如其克劳德工具）的使用。然而，总体而言，该公司收入每年增长约十倍，并计划于11月进行首次公开​​募股（IPO），一些人预计届时该公司的估值将达到创纪录的2万亿美元。

迈克尔和国防部长皮特·赫格塞斯曾在公开声明中批评过Anthropic公司。今年1月，赫格塞斯在SpaceX的一次活动上发表讲话时表示：“国防部的AI不会迎合‘觉醒’理念。它会为我们服务。我们正在制造可用于战争的武器和系统，而不是为常春藤盟校的教职工休息室开发的聊天机器人。”

2月24日，美国国防部发出最后通牒，要求Anthropic公司在2月27日前允许其人工智能技术用于所有合法用途，但遭到拒绝。2月27日，赫格塞斯宣布指示国防部将Anthropic公司列为供应链风险企业，该决定于3月3日正式公布。

下一篇报道：被盗的FBI数据揭露了员工在情报和监视工作中的角色