内容来源：https://www.theverge.com/ai-artificial-intelligence/902149/anthropic-dod-pentagon-lawsuit-supply-chain-risk-injunction

内容总结：

在与美国国防部持续数周的对峙后，人工智能公司Anthropic取得一项关键法律进展。加州北区联邦地区法院法官Rita F. Lin近日批准了该公司申请的初步禁令，暂时中止了国防部将其列入“供应链风险”名单的决定。法官在裁决中指出，国防部因Anthropic通过媒体表达对政府合同条款的质疑而实施惩罚，涉嫌违反美国宪法第一修正案所保护的言论自由。

此次纠纷源于国防部长今年1月发布的一份备忘录，要求所有人工智能服务采购合同必须在180天内加入“任何合法用途”条款。Anthropic在后续谈判中坚持两项“红线”：禁止将其AI技术用于国内大规模监控及自主致命武器系统。由于双方未能达成一致，国防部最终将这家美国本土企业列为“供应链风险”——此类认定通常仅针对与外国关联的企业。

法官在听证会上特别质疑国防部长通过社交媒体发布的“禁止所有军方合作方与Anthropic开展商业往来”的声明，指出该表述造成了广泛混乱。当被问及是否为军方提供卫生纸的承包商也会因使用Anthropic产品而被终止合同时，国防部代表未能给出明确答复。

Anthropic在诉讼文件中表示，该认定已对公司业务造成实质性损害，可能导致数百亿收入面临风险。公司发言人表示，虽然诉讼是为保护企业权益的必要举措，但仍期待与政府建设性合作，确保所有美国人受益于安全可靠的人工智能。

本案最终判决预计还需数周或数月。法官强调，其职责并非裁定AI伦理争议的是非，而是审查政府行为是否合法。这场备受关注的法律交锋，折射出美国科技企业在国家安全与商业伦理之间面临的复杂挑战。

中文翻译：

经过与五角大楼长达数周的对峙，人工智能公司Anthropic取得了一项里程碑式胜利：联邦法官批准了该公司申请的初步禁令，使其在司法程序进行期间得以暂时解除政府供应链黑名单的限制。

法官支持Anthropic诉求，暂缓执行五角大楼禁令

加州北区联邦地区法官林瑞塔在裁决书中指出：“惩罚Anthropic的行为……属于典型的违宪第一修正案报复。”该禁令将在七天后生效。

林法官在裁决书中写道：“战争部的记录显示，该部门将Anthropic列为供应链风险的原因是‘其通过媒体展现的敌对态度’。因Anthropic促使公众监督政府合同立场而实施惩罚，正是典型的违宪第一修正案报复行为。”最终判决可能需等待数周乃至数月。

Anthropic发言人丹妮尔·科恩周四声明称：“我们感谢法院迅速采取行动，并欣慰于法庭认同Anthropic的主张具有胜诉可能性。虽然本案对保护Anthropic、客户及合作伙伴确有必要，我们的重点仍是与政府建设性合作，确保所有美国人都能受益于安全可靠的人工智能。”

“此案确实触及重要争议。”林法官在周二听证会上表示，“一方面Anthropic主张其AI产品Claude不应被用于自主致命武器和国内大规模监控，坚持政府若使用其技术必须同意不用于上述用途；另一方面战争部则认为军事指挥官应自主决定AI的安全使用范围。”

她继续阐明：“我的职责并非裁决争议孰是孰非……战争部有权自主选择采购何种AI产品。包括Anthropic在内的各方都认同，该部门可自由停用Claude并寻找限制更少的AI供应商。本案的核心问题在于……政府的行为是否超越了法律边界。”

这场风波始于国防部长皮特·赫格塞斯1月9日发布的备忘录，要求180天内在所有AI服务采购合同中加入“任何合法用途”条款，涵盖Anthropic、OpenAI、xAI和谷歌等现有合作企业。Anthropic与五角大楼长达数周的谈判围绕两条“红线”展开：禁止军方将其AI用于国内大规模监控及自主致命武器（即无需人类决策即可消灭目标的AI系统）。此后事态急转直下，演变为社交媒体上的激烈指责、可能重创Anthropic业务的“供应链风险”官方认定、竞争对手趁势争夺合同，最终引发诉讼。

Anthropic在诉讼中主张，公司因受第一修正案保护的言论而遭受惩罚，要求撤销供应链风险认定。美国企业被列为供应链风险实属罕见（此前可能闻所未闻），该认定通常针对与非盟国存在潜在关联的外国企业。此事引发全国关注及两党争议，各界担忧企业若与执政当局意见相左，可能遭受过度报复。

根据法庭文件，该认定已对Anthropic业务造成显著影响：“众多外部合作伙伴联系公司……对合作要求表示困惑，担忧能否继续与Anthropic合作”“数十家企业联系公司咨询终止使用的权利”。该公司声称，根据政府对承包商与Anthropic合作的不同限制级别，可能面临数亿至数十亿美元的收入风险。

周二听证会上，双方就林法官提前一天公布的问题进行回应，焦点包括赫格塞斯是否越权发布指令及Anthropic被列为供应链风险的原因。法官在预发布问题中还质询了政府承包商因使用Anthropic技术被解约的情形，例如“若战争部承包商使用Claude代码工具为其国家安全系统编写软件，该承包商是否会因此被解约？”

林法官同时谴责战争部造成广泛混乱的赫格塞斯社交媒体帖文，该帖宣称“立即生效，所有与美军有业务往来的承包商、供应商或合作伙伴不得与Anthropic进行任何商业活动”。她在听证会上指出：“贵方此刻辩称‘虽有此言却非本意’”，并追问赫格塞斯为何发布全面禁令而非仅将Anthropic列为供应链风险。

针对法官关于“承包商在与战争部无关的业务中使用Anthropic会否被解约”的质询，战争部代表回应“据我理解确实如此”。当法官追问“若仅为战争部提供厕纸的军事承包商使用Anthropic是否会被解约”时，该代表给出肯定答复；但对于“为战争部提供非国家安全系统IT服务的承包商使用Anthropic”的情形，则未给出明确答案。

听证期间林法官引用法庭之友陈述中的“企业谋杀未遂”表述称：“虽未必是‘谋杀’，但这显然意在扼杀Anthropic。”Anthropic律师援引赫格塞斯九段式帖文强调：“该指令正在对我们造成持续且不可逆转的损害。”

国防部近期法庭文件指控，若Anthropic认为军方跨越红线，可能“在作战行动期间试图瘫痪技术或预先更改模型行为”——五角大楼称此假设情境构成“不可接受的国家安全风险”。法官的预发布问题对此提出质疑，要求提供更多证据：“现有记录如何证明Anthropic在向政府交付Claude后仍能持续访问或控制该系统，以致能够实施此类破坏行为？”

英文来源：

After Anthropic’s weeks-long standoff with the Pentagon, the company won one milestone: A judge granted Anthropic a preliminary injunction in its lawsuit, which sought to reverse its government blacklisting while the judicial process plays out.
Judge sides with Anthropic to temporarily block the Pentagon’s ban
Judge Lin wrote that ‘punishing Anthropic … is classic illegal First Amendment retaliation.’
Judge sides with Anthropic to temporarily block the Pentagon’s ban
Judge Lin wrote that ‘punishing Anthropic … is classic illegal First Amendment retaliation.’
“The Department of War’s records show that it designated Anthropic as a supply chain risk because of its ‘hostile manner through the press,’” Judge Rita F. Lin, a district judge in the northern district of California, wrote in the order, which will go into effect in seven days. “Punishing Anthropic for bringing public scrutiny to the government’s contracting position is classic illegal First Amendment retaliation.”
A final verdict could be weeks or months out.
Anthropic spokesperson Danielle Cohen said in a Thursday statement, “We’re grateful to the court for moving swiftly, and pleased they agree Anthropic is likely to succeed on the merits. While this case was necessary to protect Anthropic, our customers, and our partners, our focus remains on working productively with the government to ensure all Americans benefit from safe, reliable AI.”
“I do think this case touches on an important debate,” Judge Lin said during the Tuesday hearing. “On the one hand, Anthropic is saying that its AI product, Claude, is not safe to use for autonomous lethal weapons and domestic mass surveillance. Anthropic’s position is that if the government wants to use its technology, the government has to agree not to use it for those purposes. On the other hand the Department of War is saying that military commanders have to decide what is safe for its AI to do.”
On Tuesday, Judge Lin went on to say, “It’s not my role to decide who’s right in that debate… The Department of War decides what AI product it wants to use and buy. And everyone, including Anthropic, agrees that the Department of War is free to stop using Claude and look for a more permissive AI vendor.” She added, “I see the question in this case as being … whether the government violated the law when it went beyond that.”
It all started with a memo sent by Defense Secretary Pete Hegseth on Jan. 9, calling for “any lawful use” language to be written into any AI services procurement contract within 180 days, which would include existing contracts with companies like Anthropic, OpenAI, xAI, and Google. Anthropic’s negotiations with the Pentagon stretched on for weeks, hinging on two “red lines” that the company did not want the military to use its AI for: domestic mass surveillance and lethal autonomous weapons (or AI systems with the power to kill targets with no human involvement in the decisionmaking process). The rollercoaster series of events that followed has included a barrage of social media insults, a formal “supply chain risk” designation with the potential to significantly handicap Anthropic’s business, competing AI companies swooping in to make deals, and an ensuing lawsuit.
With its lawsuit, Anthropic argues that it was punished for speech protected under the First Amendment, and it’s seeking to reverse the supply chain risk designation.
It’s rare, and potentially even unheard of until now, for a US company to be named a supply chain risk, a designation typically reserved for non-US companies potentially linked to foreign adversaries. Anthropic’s designation as such raised eyebrows nationwide and caused bipartisan controversy due to concerns that disagreeing with a presidential administration could potentially lead to outsized retribution for a business in any sector.
Anthropic’s own business has been significantly affected by the designation, according to its court filings, which say that it has “received outreach from numerous outside partners … expressing confusion about what was required of them and concern about their ability to continue to work with Anthropic” and that “dozens of companies have contacted Anthropic” for guidance or information about their rights to terminate usage. Depending on the level to which the government prohibits its contractors’ work with Anthropic, the company alleged that revenue adding up to between hundreds of millions and multiple billions could be at risk.
During Tuesday’s hearing, both companies had a chance to respond to Judge Lin’s questions, which were released in a document the day prior and hinged on matters like whether Hegseth lacked authority to issue certain directives and why Anthropic was named a supply chain risk. The judge also asked, in her pre-released questions, about the circumstances under which a government contractor could face termination for using Anthropic’s technology in their work — for instance, “if a contractor for the Department uses Claude Code as a tool to write software for the Department’s national security systems, would that contractor face termination as a result?”
On Tuesday, the judge also seemed to admonish the Department of War for Hegseth’s X post that caused a lot of widespread confusion per Anthropic’s earlier court filings, stating that “effective immediately, no contractor, supplier, or partner that does business with the United States military may conduct any commercial activity with Anthropic.”
“You’re standing here saying, ‘We said it but we didn’t really mean it,’” Judge Lin said during the hearing, later pressing on the question of why Hegseth wrote the above barring contractors from working with Anthropic instead of just simply designating Anthropic as a supply chain risk.
In a series of questions on Tuesday, Judge Lin asked whether the Department of War plans to terminate contractors on the basis of their work with Anthropic if it’s separate from their work with the department, and a representative for the Department of War responded, “That is my understanding.”
Judge Lin asked, “Let’s say I’m a military contractor. I don’t provide IT to the military. I provide toilet paper to the military. I’m not going to be terminated for using Anthropic — is that accurate?” The representative for the Department of War responded, “For non-DoW work, that is my understanding.” But when the judge asked whether a military contractor providing IT services to the Department of War, but not for national security systems, could be terminated for using Anthropic, the representative for the Department of War did not give a concrete answer.
During the hearing, Judge Lin cited one of the amicus briefs, which she said used the term “attempted corporate murder.” She said, “I don’t know if it’s ‘murder,’ but it looks like an attempt to cripple Anthropic.”
“We are continuing to be irreparably injured by this directive,” a lawyer for Anthropic said during the hearing, citing Hegseth’s nine-paragraph X post.
In a recent court filing, the Department of Defense alleged that Anthropic could ostensibly “attempt to disable its technology or preemptively alter the behavior of its model either before or during ongoing warfighting operations” in the event it felt the military was crossing its red lines — a theoretical situation that the Pentagon said it deemed an “unacceptable risk to national security.” The judge’s pre-released questions seem to challenge that statement, or at least request more information on it, stating, “What evidence in the record shows that Anthropic had ongoing access to or control over Claude after delivering it to the government, such that Anthropic could engage in such acts of sabotage or subversion?”