# Anthropic Said No to Autonomous Weapons. The U.S. Is Fighting Back. Source: https://www.youtube.com/watch?v=QhQp_z16I_A Recap page: https://rapidrecap.app/video/QhQp_z16I_A Generated: 2026-02-20T12:33:07.477+00:00 --- ## Quick Overview Anthropic is in a dispute with the Pentagon over a $200 million contract, threatening to walk away because the government wants to use Claude for surveillance and autonomous weapon operations, which violates Anthropic's stated ethical principles, leading to a political battle over AI safety and governmental access to powerful models. **Key Points:** - Anthropic is engaged in a dispute with the Pentagon over a $200 million contract, threatening to drop it due to ethical concerns regarding AI use in warfare. - The dispute centers on the Pentagon's desire to use Anthropic's Claude model for surveillance and autonomous weapon operations, which Anthropic opposes. - Anthropic demanded two changes to the contract: removing clauses that would allow the US military to use Claude for mass domestic surveillance or autonomous killing operations without a human in the loop. - Three of the four AI labs contracted by the Pentagon (OpenAI, XAI, and Google) have signed an 'all lawful uses' contract, unlike Anthropic. - The Pentagon is reportedly willing to take a revenue hit and potentially lose the contract rather than yield on its AI usage principles. - The conflict highlights the growing political battle over AI safety and governmental access to advanced AI technologies, drawing parallels to past political clashes. - If Anthropic walks away, the risk to the company's supply chain regarding access to US military contracts and potentially other government work is significant. ![Screenshot at 00:40: A New York Times headline graphic is displayed, reading "Defense Dept. and Anthropic Square Off in Dispute Over A.I. Safety," illustrating the core conflict being discussed.](https://ss.rapidrecap.app/screens/QhQp_z16I_A/00-00-40.jpg) **Context:** The discussion centers on a significant conflict between the AI company Anthropic and the US Department of Defense (the Pentagon) regarding a potential $200 million contract. Anthropic, which positions itself as a safety-focused AI lab, is resisting specific contract terms that would allow the Pentagon to deploy its Claude model for military applications deemed unethical by the company, such as autonomous weapon operations and domestic surveillance. ## Detailed Analysis The podcast hosts, Kevin Roose and Casey Newton, discuss the ongoing dispute between Anthropic and the Pentagon over a $200 million contract for the use of Anthropic's Claude AI model. Kevin Roose explains that Anthropic's boyfriend works at Anthropic, and he works at the New York Times, which covers Open AI, Microsoft, and perplexity. Anthropic is reportedly upset over the terms of the contract, threatening to drop it. The core issue is Anthropic's refusal to allow its models to be used for mass domestic surveillance or autonomous lethal operations without a human in the loop, which conflicts with the Pentagon's perceived needs. Roose notes that three other major AI contractors—OpenAI, XAI, and Google—have signed a contract allowing 'all lawful uses,' unlike Anthropic. The Pentagon is reportedly prepared to cancel the contract rather than change its stance, considering the risk of supply chain issues for Anthropic. The hosts suggest that the fight is less about abstract morality and more about capability and political alignment, noting that the Trump administration previously pushed for a 10-year moratorium on military use of AI, which Anthropic executives were reportedly aligned with. The conflict escalates because the military wants to use Claude for tasks like capturing the President of Venezuela, which Anthropic explicitly wants to prevent. The hosts conclude that the dispute is a significant political battle over AI regulation and control, with Anthropic standing firm on its safety principles despite the financial implications. ### Pentagon-Anthropic Dispute - Ongoing conflict over a $200 million contract - Anthropic threatens to drop the deal due to ethical concerns over military use - Pentagon wants unrestricted use, Anthropic wants guardrails. ### Contract Terms Conflict - Anthropic demands removal of clauses for mass domestic surveillance and autonomous killing bots - Other labs (OpenAI, XAI, Google) signed 'all lawful uses' contracts. ### The Stakes - Pentagon might cancel the contract rather than yield, potentially impacting Anthropic's access to government work - The conflict is framed as a fight over political alignment and capabilities, not just ethics. ### Dario Amodei's Stance - Amodei (Anthropic CEO) highlighted risks like surveillance and autonomous weapons in his essay, "The Adolescence of Technology," indicating a principled stand. ### Political Context - The conflict mirrors past friction between tech companies and administrations (like Trump's) regarding military AI use and export controls. ![Screenshot at 00:01: The two hosts, Kevin Roose \(left\) and Casey Newton \(right\), are seated at a round table with microphones, beginning their discussion.](https://ss.rapidrecap.app/screens/QhQp_z16I_A/00-00-01.jpg) ![Screenshot at 00:40: A New York Times headline graphic is displayed, reading "Defense Dept. and Anthropic Square Off in Dispute Over A.I. Safety," illustrating the core conflict being discussed.](https://ss.rapidrecap.app/screens/QhQp_z16I_A/00-00-40.jpg) ![Screenshot at 02:01: A graphic slide appears showing the title of Dario Amodei's essay: "The Adolescence of Technology: Confronting and Overcoming the Risks of Powerful AI."](https://ss.rapidrecap.app/screens/QhQp_z16I_A/00-02-01.jpg) ![Screenshot at 10:00: A slide displaying the title of Dario Amodei's essay, "The Adolescence of Technology," is shown against a dark background.](https://ss.rapidrecap.app/screens/QhQp_z16I_A/00-10-00.jpg) ![Screenshot at 17:04: A New York Times article headline is displayed: "Homeland Security Wants Social Media Sites to Expose Anti-ICE Accounts," showing a relevant parallel case of government overreach concerns.](https://ss.rapidrecap.app/screens/QhQp_z16I_A/00-17-04.jpg)