# The Hidden Cost of OpenAI’s Pentagon Deal? Trust. Source: https://www.youtube.com/watch?v=tnv3kIGU8go Recap page: https://rapidrecap.app/video/tnv3kIGU8go Generated: 2026-03-06T15:32:53.285+00:00 --- ## Quick Overview The core issue stemming from OpenAI's Pentagon deal is the perceived lack of trust and transparency, especially as the company now faces public outcry and internal dissent over its relationship with military entities, highlighted by the revelation that Anthropic's Claude AI is reportedly being used by the U.S. military in the Iran conflict, suggesting that the stated commitment to preventing harmful military applications is not being strictly enforced, leading to calls for legislative action. **Key Points:** - The Pentagon officially designated Anthropic a 'supply chain risk' due to its work with the U.S. government, which could prevent the startup from doing business with the U.S. government. - Sam Altman announced OpenAI's agreement with the Pentagon, which included prohibitions on using the technology for domestic surveillance or autonomous weapons systems, but this was later criticized as insufficient. - Following the Pentagon's classification of Anthropic as a risk, news surfaced that Anthropic's Claude AI is allegedly being used by the U.S. military in the Iran war. - Prediction markets like Kalshi and Polymarket saw high trading volume and high probability bets regarding the Iran conflict and the potential ouster of Iran's Supreme Leader, Ali Khamenei. - Kalshi voided some bets related to Khamenei's ouster because they were 'directly tied to death,' leading to user outcry over unclear rules regarding profiting from death. - The discussion suggests that if Anthropic continues its relationship with the Pentagon, it risks alienating its internal talent who prioritize safety and non-military applications, contrasting with the Silicon Valley ethos of worker empowerment. - The existence of these prediction markets trading on geopolitical events like wars and assassinations demonstrates a market for information that is ethically questionable, regardless of legality. ![Screenshot at 00:16: The New York Times article graphic detailing the Pentagon's official notification to Anthropic that it is a 'supply chain risk,' which could prevent future government business.](https://ss.rapidrecap.app/screens/tnv3kIGU8go/00-00-16.jpg) **Context:** The discussion centers on the fallout from recent controversies involving major AI companies, specifically OpenAI and Anthropic, and their contracts with the U.S. Department of Defense (Pentagon). Kevin Roose and Casey Newton analyze the erosion of public and employee trust in these AI firms following the revelation that Anthropic's Claude AI is allegedly being used by the U.S. military in foreign conflicts, juxtaposed against the companies' stated ethical guidelines. ## Detailed Analysis The conversation unpacks the high-stakes drama surrounding OpenAI and Anthropic's entanglement with the Pentagon, arguing that these deals fundamentally undermine the trust these companies rely upon. Kevin Roose notes the initial backlash against OpenAI's Pentagon deal, which Sam Altman tried to mitigate by stating prohibitions on surveillance and autonomous weapons use, but this clarity proved insufficient when it was revealed that Anthropic's Claude AI was reportedly being used by the U.S. military in the Iran conflict. This contradiction fueled public anger, evidenced by the 'QuitGPT' movement. Casey Newton points out that the internal dissent at Anthropic, including high-profile departures like Max Schwarz, stems from employees who value the company's original non-profit-like mission focused on beneficial AI, contrasting sharply with the government contracting work. Furthermore, the discussion touches on the existence of prediction markets like Kalshi and Polymarket, where users were betting heavily on geopolitical events like the Iran conflict and the potential death of Ali Khamenei. Kalshi's subsequent decision to void bets tied to death, citing unclear rules, further fueled public distrust and hypocrisy claims. The hosts conclude that the government's desire for powerful AI is clashing with the industry's attempts to maintain an ethical public image, suggesting that the government's approach—either through direct control via nationalization or subtle influence via contracts—is creating a precarious situation for AI development, especially given the rapid advancement of models like GPT-6 and GPT-7. ### Pentagon vs. Anthropic - Pentagon formally designated Anthropic a 'supply chain risk' (00:13), leading to threats of losing government business, while Anthropic's competitor OpenAI secured a deal with the Pentagon that included stated ethical limitations (00:50). ### Public and Employee Backlash - Public outcry led to the 'QuitGPT' boycott movement against OpenAI (00:55), and Anthropic faced internal dissent from employees who felt the company was abandoning its safety-first mission (05:09, 08:58). ### Prediction Markets and Geopolitics - Prediction markets like Kalshi and Polymarket saw significant activity betting on geopolitical events, such as the Iran conflict (2:16, 10:58) and Khamenei's ouster (25:55), highlighting a market for morbid speculation. ### Kalshi's Rule Enforcement - Kalshi voided bets tied to death (25:57), citing rules against profiting from death, which users criticized as inconsistent and unclear (26:01). ### The Risk of Government Control - The hosts argue that government involvement, whether overt or through subtle contract pressure, forces AI companies into a state of constant PR management, potentially leading to nationalization or excessive control over technology development (04:44, 07:01, 17:04). ### Future Trajectory - The hosts suggest that if the government continues to push for military applications, it risks losing key talent who prefer open, safety-oriented development, creating a dangerous dichotomy between safety and state interests (08:38, 13:32). ![Screenshot at 00:16: The New York Times article graphic detailing the Pentagon's official notification to Anthropic that it is a 'supply chain risk,' which could prevent future government business.](https://ss.rapidrecap.app/screens/tnv3kIGU8go/00-00-16.jpg) ![Screenshot at 01:05: The New York Times article showing OpenAI amending its A.I. deal with the Pentagon to include additional protections against mass surveillance.](https://ss.rapidrecap.app/screens/tnv3kIGU8go/00-01-05.jpg) ![Screenshot at 01:56: A TechCrunch headline reporting that ChatGPT uninstalls surged by 295% following the DoD deal.](https://ss.rapidrecap.app/screens/tnv3kIGU8go/00-01-56.jpg) ![Screenshot at 25:55: A CoinDesk article showing Polymarket shelving nuclear detonation markets following public outcry.](https://ss.rapidrecap.app/screens/tnv3kIGU8go/00-25-55.jpg)