五角大楼开除Anthropic为小型AI竞争对手打开大门


2026-04-09 10:03:34 UTC / 路透社

作者:迈克·斯通

2026年4月9日 美国东部时间上午10:03 更新于1小时前

节点运行失败

2026年3月1日制作的示意图中展示了美国战争部与Anthropic的标志。路透社/达多·鲁维奇/示意图/档案照片/档案照片 购买授权,打开新标签页

华盛顿4月9日(路透社)——在美国国防部与其曾经青睐的人工智能供应商Anthropic关系恶化,凸显出军方需要实现多元化并增加人工智能供应商数量后,小型国防工业人工智能初创企业突然接到了来自将军们、作战指挥官和财力雄厚的投资者的来电。

在国防部与Anthic的矛盾公开化并导致该公司被逐出美军后的几周里,专注国防领域的新人工智能公司如Smack Technologies和EdgeRunner AI表示,他们经历了数月前难以想象的兴趣转变。他们收到了大量关于潜在合同的接洽和会面请求,此前毫无兴趣的投资者也主动找上门来。

通过《每日案卷》通讯将最新法律新闻直接发送到您的收件箱,开启您的清晨。在此处订阅。

广告 · 滚动继续阅读

五角大楼对其顶级人工智能供应商Anthropic日益增长的敌意,为小型竞争对手打开了机会之门,这些公司长期以来一直渴望进入这个全球最赚钱的政府承包商市场。一份国防合同可以带来与美国其他政府部门的更多业务,同时也能向潜在商业客户传递信任与安全的积极信号。

“自从Anthropic被认定为供应链风险以来,我们看到客户和政府对部署人工智能解决方案的需求大幅增加,”Second Front公司首席执行官泰勒·斯韦特说道,该公司帮助科技公司满足在五角大楼安全网络上运营所需的合规要求。“随着Anthic事件的爆发,五角大楼转而寻求我们的客户快速部署解决方案,我们的客户也开始向我们求助。”

广告 · 滚动继续阅读

自3月份五角大楼将Anthropic的产品认定为“供应链风险”,双方陷入诉讼以来,军方对Smack Technologies等人工智能初创企业的兴趣与日俱增。总部位于加州埃尔塞贡多、拥有19名员工的初创公司联合创始人兼首席执行官安德鲁·马科夫表示:“我们希望获得更多产品,希望进行演示,谈谈如何加快推进速度。”3月底,一名法官暂时阻止了五角大楼对Anthropic的黑名单制裁。

EdgeRunner AI联合创始人兼首席执行官泰勒·萨尔茨曼也描述了类似的经历。他的公司等待太空军合同通过五角大楼采购流程已经超过一年。但在Anthropic事件公开后的几周内,合同就签署了。“我无法证明Anthropic的风波加速了这一进程,”萨尔茨曼说,“但我私下怀疑确实如此。”

“五角大楼将继续通过与所有保密级别行业合作伙伴的牢固合作,快速向前线作战人员部署前沿人工智能能力,”一名五角大楼官员表示。

一名五角大楼技术专家此前曾告诉路透社,与Anthropic的决裂,以及意识到国防部严重依赖单一人工智能供应商,迫使该部门实现人工智能供应商多元化。

SMACK与海军陆战队的合同加速推进

对于Smack而言,Anthropic事件后加速进程的最明显例子涉及海军陆战队。该公司于2025年3月赢得了海军陆战队的合同,并于10月交付了成功的原型软件——该软件可将通常需要数月的作战规划流程压缩至约15分钟。

尽管原型取得成功,但推进势头陷入停滞。全面生产原计划在2027财年——也就是最早2027年10月——纳入预算。在2025年假期期间直至2026年初,项目没有明确方向。

随后Anthropic事件爆发。几周内,Smack就受邀与海军陆战队举行多次会议,核心问题只有一个:今年能以多快的速度投入生产。马科夫表示,军方“给出了非常明确的指导、行动和动力”,推动原型在2026年做好投入作战行动的准备——这比原计划提前了一年多。

这种转变不仅限于海军陆战队。Smack与海军和空军也有合同,马科夫表示,美国特种作战司令部等部门几乎立刻就表达了兴趣。

EdgeRunner目前正在与陆军特种部队小组合作,并已获得太空军的合同,该公司表示海军也大幅加快了合作节奏。原本每两周或每月举行一次的会议,现在每周多次召开。

EdgeRunner和Smack目前都在竞相将其系统提升至更高的安全保密级别——这是进入最具作战意义的应用场景和最大规模军事合同的准入门槛。

EdgeRunner表示,军方告知该公司,可以在三个月内达到IL-6安全级别,这一级别允许访问机密和绝密数据。萨尔茨曼称,这一时间表“非同寻常”,因为正常流程通常需要18个月或更长时间。他表示,这种加速一方面是由于五角大楼领导层推动打破采购官僚主义的压力,另一方面也是由于Anthropic事件给国防部的人工智能战略注入了紧迫感。

迈克·斯通在华盛顿报道 克里斯·桑德斯和马修·刘易斯编辑

我们的标准:路透社信任原则,打开新标签页

Pentagon’s ouster of Anthropic opens doors for small AI rivals

2026-04-09 10:03:34 UTC / Reuters

By Mike Stone

April 9, 2026 10:03 AM UTC Updated 1 hour ago

节点运行失败

U.S. Department of War and Anthropic logos are seen in this illustration created on March 1, 2026. REUTERS/Dado Ruvic/Illustration/File Photo/File Photo Purchase Licensing Rights, opens new tab

WASHINGTON, April 9 (Reuters) – Small defense industry artificial intelligence startups are suddenly fielding calls from generals, combatant commanders and deep-pocketed investors, after the souring relationship between the Pentagon and its ​once-favored AI vendor, Anthropic, reinforced the need to diversify and increase the number of AI providers for the military.

In the weeks since the Department of Defense’s troubled ‌relationship with Anthropic burst into public view and led to the company being kicked out of the U.S. military, new defense-focused AI companies like Smack Technologies and EdgeRunner AI say they have experienced a shift in interest that would have been unimaginable just months ago. They have received a surge of overtures about possible contracts and meeting requests and been approached by investors who previously showed no interest.

Jumpstart your morning with the latest legal news delivered straight to your inbox from The Daily Docket newsletter. Sign up here.

Advertisement · Scroll to continue

The Pentagon’s growing animosity toward its top AI provider, Anthropic, has opened ​up opportunities for smaller rivals, who have long sought a foot in the door to the most lucrative government contractor in the world. A defense contract can lead to more ​business with other branches of the U.S. government, and is a useful signal of trust and safety for potential commercial clients.

“We’ve seen a massive ⁠increase in demand from customers and the government to get AI solutions fielded since Anthropic was declared a supply-chain risk,” said Tyler Sweatt, CEO of Second Front, a company that helps technology firms ​meet the requirements needed to operate on secure Pentagon networks. “Our customers are turning to us as the Pentagon turns to them to deploy quickly in the wake of the Anthropic blowup.”

Advertisement · Scroll to continue

Since the Pentagon deemed ​Anthropic’s products a “supply-chain risk” in March and the two sides became embroiled in a lawsuit, the military has expressed increasing interest in AI startups like Smack Technologies, saying, “We want more, we want demos, let’s talk about how we can move faster,” said Andrew Markoff, co-founder and chief executive of the 19-person startup based in El Segundo, California. In late March, a judge temporarily blocked the Pentagon’s blacklisting of Anthropic.

Tyler Saltsman, co-founder and chief executive of EdgeRunner AI, described ​a similar experience. His company had been waiting more than a year for a Space Force contract to clear the Pentagon’s procurement machinery. It was signed within weeks of the Anthropic situation breaking ​into the open. “I can’t prove that the Anthropic drama sped this up,” Saltsman said, “but I have a sneaky suspicion it did.”

“The Pentagon will continue to rapidly deploy frontier AI capabilities to the warfighter through strong industry partnerships ‌across all ⁠classification levels,” a Pentagon official said.

One Pentagon technologist has previously told Reuters that the falling-out with Anthropic, and the realization that the Defense Department was heavily dependent on one AI provider, forced the department to diversify AI providers.

SMACK’S MARINE CORPS CONTRACT SPEEDS UP

For Smack, the clearest example of the post-Anthropic acceleration involves the Marine Corps. The company won a contract with the Marine Corps in March 2025 and delivered a successful prototype by October — software that compresses what is normally a months-long operational planning process into roughly 15 minutes.

Despite the successful prototype, momentum stalled. Full production had been budgeted for fiscal year ​2027 — meaning October 2027 at the earliest. Through ​the 2025 holiday period and into early 2026, ⁠there was no clear direction.

Then the Anthropic uproar occurred. Within weeks, Smack was invited to multiple meetings with the Marine Corps focused on a single question: how fast can this move into production this year? Markoff said there was “very specific guidance and movement and energy” toward getting the prototype ready ​for combat operations in 2026 — an acceleration of more than a year.

The shift extended beyond the Marines. Smack holds contracts with the Navy and ​Air Force, and Markoff ⁠said interest came in nearly immediately from U.S. Special Operations Command, and others.

EdgeRunner, which is deploying with the Army Special Forces groups and has received a contract with the Space Force, said the Navy has also dramatically sped up engagement. Meetings that had been biweekly or monthly are now happening multiple times a week.

Both EdgeRunner and Smack are now racing to get their systems operating at higher security classification levels — the ⁠gateway to the ​most operationally significant use cases and the largest military contracts.

EdgeRunner said the military has told the company it can get ​to IL-6, a security designation enabling access to secret and top-secret data, within three months — a timeline Saltsman described as remarkable, given that the process normally takes 18 months or longer. The acceleration, he said, is being driven partly by pressure ​from Pentagon leadership to cut through procurement bureaucracy, and partly by the urgency the Anthropic situation has injected into the department’s AI strategy.

Reporting by Mike Stone in Washington Editing by Chris Sanders and Matthew Lewis

Our Standards: The Thomson Reuters Trust Principles., opens new tab

评论

发表回复

您的邮箱地址不会被公开。 必填项已用 * 标注