# OpenAI Vs. Anthropic: How the Pentagon Picked Its Partner Source: https://www.youtube.com/watch?v=Zj35mEtwUvY Recap page: https://rapidrecap.app/video/Zj35mEtwUvY Generated: 2026-03-01T01:02:49.527+00:00 --- ## Quick Overview The Pentagon ultimately favored OpenAI over Anthropic in securing a contract for deploying AI models in classified networks because Anthropic's stated red lines regarding mass surveillance and autonomous weapons, reinforced by a recent public statement from CEO Dario Amodei, were deemed less palatable or compatible with the Pentagon's evolving needs compared to OpenAI's perceived willingness to engage, despite the political fallout from Trump's intervention. **Key Points:** - Anthropic CEO Dario Amodei publicly stated two red lines: prohibiting mass surveillance and fully autonomous weapons, leading to perceived conflict with the Pentagon. - The Pentagon signed a deal with OpenAI to deploy their models in classified networks, suggesting a preference for a partner willing to engage on sensitive use cases. - Dean W. Ball (NYT) noted that the explicit red lines from Anthropic were seen by some as a political maneuver to appear morally superior to Big Tech rivals. - The conflict escalated when former Under Secretary of War Emil Michael posted on Truth Social, criticizing Amodei and directing federal agencies to cease using Anthropic technology. - Anthropic's previous agreement with the Pentagon included safeguards that prohibited applications like autonomous weapons and mass surveillance, which the Pentagon then allegedly rejected. - The situation suggests that the Pentagon prioritized a partner willing to work within established (or newly defined) parameters, even if those parameters were contentious, over a company drawing explicit legal or moral lines. ![Screenshot at 00:19: Casey Newton and Kevin Roose discuss the 48-hour period involving the Pentagon, Anthropic, and OpenAI, where the dispute over AI use cases came to a head.](https://ss.rapidrecap.app/screens/Zj35mEtwUvY/00-00-19.jpg) **Context:** This discussion revolves around the fallout from a highly publicized dispute between the US Department of Defense (Pentagon) and two leading AI safety companies, Anthropic and OpenAI, concerning contracts for deploying AI models within classified government networks. The conflict intensified following a public statement by Anthropic CEO Dario Amodei detailing ethical red lines that conflicted with certain military applications, prompting a sharp public rebuke from a former Trump administration official and highlighting the ongoing tension between corporate ethics and national security requirements in the AI sector. ## Detailed Analysis The discussion centers on the fallout from the Pentagon's decision to partner with OpenAI over Anthropic for deploying AI models in classified networks. Kevin Roose notes the extremely fast-paced nature of the events following Anthropic CEO Dario Amodei’s public statement outlining two key red lines: prohibiting mass surveillance and fully autonomous weapons applications. Casey Newton suggests that this statement, which seemed designed to draw clear ethical boundaries, may have been perceived by the Pentagon as either naive or politically motivated, especially since the Pentagon had previously agreed to similar contractual safeguards that Anthropic was now publicly emphasizing. The situation escalated when former Under Secretary of War Emil Michael posted on Truth Social attacking Amodei and ordering a halt to all federal use of Anthropic’s technology, deeming Anthropic incompatible with American principles. Roose points out that the disagreement boils down to whether the companies' stated ethical boundaries (like prohibiting mass surveillance) are genuine moral stances or just political posturing, especially when they conflict with the government's perceived needs. Ultimately, the Pentagon's selection of OpenAI suggests that OpenAI's willingness to engage on the contested uses outweighed Anthropic's stated refusal, leading to a severe strain in the relationship between Anthropic and the DoD. ### The Pentagon's AI Partner Selection - The Pentagon chose OpenAI over Anthropic for classified network AI deployment - This decision followed public conflict over Anthropic's ethical red lines - Open AI's willingness to engage on sensitive use cases likely favored them over Anthropic's stated prohibitions. ### Anthropic's Stance and Conflict Escalation - Dario Amodei's public statement outlined red lines against mass surveillance and autonomous weapons - Former Under Secretary of War Emil Michael responded aggressively on Truth Social, ordering agencies to cease Anthropic use - Michael suggested Anthropic's stance was politically motivated and signaled a permanent alteration of the relationship. ### Implications and Future Uncertainty - The conflict highlights the tension between corporate ethics and military needs in AI deployment - The specific details of the contracts regarding surveillance and autonomous weapons remain largely unknown - The speakers note the high stakes for the AI industry in setting precedents for government contracts. ![Screenshot at 00:05: Kevin Roose describes the preceding 48-hour period in the AI industry as 'very crazy'.](https://ss.rapidrecap.app/screens/Zj35mEtwUvY/00-00-05.jpg) ![Screenshot at 00:25: Casey Newton summarizes the events, noting the rapid sequence of decisions and statements.](https://ss.rapidrecap.app/screens/Zj35mEtwUvY/00-00-25.jpg) ![Screenshot at 01:19: The screen displays the headline of a New York Times article: 'Trump Orders Government to Stop Using Anthropic After Pentagon Standoff'.](https://ss.rapidrecap.app/screens/Zj35mEtwUvY/00-01-19.jpg) ![Screenshot at 03:08: A screenshot of the statement released by Anthropic CEO Dario Amodei on Anthropic's website regarding discussions with the Department of War.](https://ss.rapidrecap.app/screens/Zj35mEtwUvY/00-03-08.jpg) ![Screenshot at 06:37: A screenshot of the Truth Social post by Under Secretary of War Emil Michael attacking Dario Amodei and Anthropic's stance.](https://ss.rapidrecap.app/screens/Zj35mEtwUvY/00-06-37.jpg)