美国国防部停止使用Anthropic AI工具

美国国防部已宣布停止使用Anthropic公司开发的AI工具。1国防部长Pete Hegseth曾在2月以"国家安全供应链风险"为由将Anthropic列入黑名单，并设定8月底前完成停用。1据BBC了解，直至最近一周，Anthropic的Claude模型仍在国防部的Maven Smart System中被广泛用于研究、分析、情报收集和对伊朗军事行动。1

这一冲突源于Anthropic拒绝移除Claude的安全防护措施。1Anthropic表示做出此决定是因为担忧大规模监控和自主武器的风险。1该公司随后对特朗普政府提起诉讼，称这一举措"前所未有且非法"。1Claude已被深度集成在由Palantir公司运营的Maven系统中，用于情报分析和目标识别。1

国防部已与Google、xAI和OpenAI等公司签订合同。1特朗普上周与Anthropic首席执行官Dario Amodei在白宫进行了两次会面。1

The U.S. Department of Defense has discontinued its use of artificial intelligence tools developed by Anthropic, according to reporting by the BBC.1 Defense Secretary Pete Hegseth blacklisted the company in February on grounds of national security supply chain risks, setting a deadline of August 31 for cessation of all systems using Anthropic's technology.1 However, the BBC learned that Anthropic's Claude model remained in active deployment within the Pentagon's Maven Smart System as recently as the previous week, where it was being utilized for research, analysis, intelligence gathering, and assessments of Iranian military operations.1

The conflict between Anthropic and the Department of Defense escalated over the company's refusal to remove safety protections from its Claude model.1 Anthropic maintained that such removal posed risks related to large-scale surveillance and autonomous weapons systems, characterizing the Pentagon's directive as "unprecedented and unlawful."1 In response, Anthropic has filed a lawsuit against the Trump administration.1 Meanwhile, the Pentagon has established contracts with alternative artificial intelligence providers including Google, xAI, and OpenAI.1 The dispute drew additional attention when President Trump met with Anthropic Chief Executive Officer Dario Amodei at the White House twice in recent days.1