Anthropic defies Pentagon's $200M ultimatum over AI weapon use

[00:00] Evelyn Hartwell: From Neural Newscast, I'm Evelyn Hartwell. [00:03] Frederick Moore: And I'm Frederick Moore. [00:04] Evelyn Hartwell: Today, we are covering the intensifying standoff between AI startup Anthropic and the Pentagon, [00:12] Evelyn Hartwell: over a $200 million contract. [00:15] Frederick Moore: We will examine why CEO Dario Amodai is refusing military demands [00:21] Frederick Moore: and the potential consequences of being labeled a national supply chain risk. [00:26] Evelyn Hartwell: Anthropic is officially rejecting the Pentagon's latest offer to modify their existing contract regarding the Claude AI model. [00:36] Frederick Moore: Defense Secretary Pete Hegseth issued a direct ultimatum to AmoDy on Tuesday. [00:42] Frederick Moore: He demanded the AI be available for all lawful purposes. [00:46] Evelyn Hartwell: If Anthropic does not comply by Friday at 5.01 p.m., the military plans to cancel the $200 million partnership. [00:56] Frederick Moore: The disagreement centers on specific restrictions Anthropic places on Claude. [01:01] Frederick Moore: The model is currently the only AI on the military's classified networks. [01:07] Evelyn Hartwell: Amadai argues that today's technology is not reliable enough to power fully autonomous weapons or conduct mass surveillance safely. [01:17] Frederick Moore: He wrote in a blog post today that while he believes in defending democracy, certain use cases simply undermine democratic values. [01:27] Evelyn Hartwell: Anthropic claims the Pentagon's proposed compromises included legalese that would let the government disregard safety safeguards at will. [01:37] Frederick Moore: The company's internal alignment team is standing firm. [01:41] Frederick Moore: Staffers are publicly supporting this refusal to budge. [01:44] Evelyn Hartwell: Turning now to the Pentagon's perspective, officials are framing this as a direct threat to American readiness against adversaries like China. [01:53] Frederick Moore: Under Secretary Emil Michael was blunt on social media. [01:58] Frederick Moore: He accused Amadeh of having a God complex and trying to control the military. [02:04] Evelyn Hartwell: Michael told reporters that the military will always adhere to the law. [02:08] Evelyn Hartwell: He said they cannot let a private firm dictate defense policy. [02:12] Frederick Moore: He emphasized that no company will be allowed to remove software from the department [02:18] Frederick Moore: until a viable alternative is ready. [02:21] Evelyn Hartwell: The military has even invited Anthropic to join its AI ethics board. [02:26] Evelyn Hartwell: However, this gesture has not resolved the fundamental dispute. [02:30] Frederick Moore: The Pentagon argues that the restrictions Anthropic wants [02:35] Frederick Moore: are already covered by existing federal laws and internal department policies. [02:40] Evelyn Hartwell: Meanwhile, the consequences for Anthropic could extend far beyond the loss of this single $200 million contract. [02:50] Frederick Moore: The Pentagon is threatening to designate Anthropic as a supply chain risk. [02:55] Frederick Moore: This classification is usually reserved for hostile foreign companies. [03:00] Evelyn Hartwell: There are also reports that the administration might invoke the Defense Production Act [03:05] Evelyn Hartwell: to force the company to provide the services. [03:08] Frederick Moore: This would be a drastic escalation. [03:11] Frederick Moore: It essentially treats AI software as a critical resource that the government can seize in emergencies. [03:18] Evelyn Hartwell: As of this afternoon, Anthropic has not shown signs of reversing its position before tomorrow's deadline. [03:26] Frederick Moore: We will be watching closely as the Friday deadline passes to see if the military follows through on its threats. [03:34] Evelyn Hartwell: I'm Evelyn Hartwell. [03:35] Frederick Moore: And I'm Frederick Moore. Neural Newscast is AI-assisted, human-reviewed. [03:41] Frederick Moore: View our AI transparency policy at neuralnewscast.com. [03:46] Announcer: Neural Newscast uses artificial intelligence in content creation [03:50] Announcer: with human editorial review prior to publication. [03:53] Announcer: While we strive for factual, unbiased reporting, [03:56] Announcer: AI-assisted content may occasionally contain errors. [03:59] Announcer: Verify critical information with trusted sources. [04:02] Announcer: Learn more at neuralnewscast.com. [04:05] Announcer: you