Anthropic rejects latest Pentagon offer: ‘We cannot in good conscience accede to their request’

Anthropic has firmly declined the Pentagon's recent offer to modify their existing contract, citing significant concerns regarding the potential misuse of AI for mass surveillance and fully autonomous weaponry. This disagreement highlights the tension between the defense establishment and the AI company regarding the limitations placed on Claude, the pioneering AI system integrated into the military's classified operations. During a conversation on Tuesday, Defense Secretary Pete Hegseth warned Anthropic CEO Dario Amodei that failure to permit the use of their AI model for 'all lawful purposes' could lead to the cancellation of a substantial $200 million contract. Additionally, Hegseth indicated that Anthropic would be classified as a 'supply chain risk,' a label typically associated with entities linked to foreign adversaries. In a statement, Anthropic expressed that the Pentagon's revised terms, while presented as a compromise, included complex legal language that could potentially undermine the very safeguards the company sought to protect. In a comprehensive blog post on Thursday, Amodei articulated his deep commitment to leveraging AI in defense of the United States and other democratic nations, emphasizing the need to counter autocratic threats. Amodei acknowledged the Pentagon's authority in military decision-making but maintained that in specific scenarios, the deployment of AI could actually jeopardize democratic values. He asserted that issues like mass surveillance and autonomous weapons are beyond the current technological capabilities to manage safely and effectively. Despite the Pentagon's implied threats, Amodei reiterated, "we cannot in good conscience accede to their request." The Pentagon has yet to provide a response regarding this ongoing dispute.