In late February 2026, the AI industry could be described as “war clouds gathering.” Within just 24 hours, the U.S. Department of Defense (the Pentagon; now also called the War Department) issued major decisions targeting two AI giants—OpenAI and Anthropic—triggering a seismic tremor across the tech industry. Not only that, employees from Google and OpenAI also rarely joined hands to jointly express support for Anthropic. This power struggle between the AI industry and military institutions is redefining the boundaries of AI safety.
OpenAI Reaches Agreement with the Pentagon; Model to Be Deployed on U.S. Classified Networks
OpenAI founder Sam Altman announced earlier on social media that OpenAI has reached an agreement with the US Department of Defense (referred to as the “War Department”) to deploy OpenAI’s AI models within the Defense Department’s classified networks. Altman emphasized that the safety and widespread benefits of artificial intelligence are central to OpenAI’s mission, and listed two of the most important safety principles:
- Ban large-scale domestic surveillance
- The use of force must remain under human responsibility, including autonomous weapons systems.
Altman said the Department of Defense endorses these principles and has incorporated them into laws and policies, and OpenAI has also written these principles into its agreements. Additionally, OpenAI will build technical security safeguards to ensure models operate as intended, while deploying feature development engineers to assist with model operations and ensure safety, and will only deploy in cloud networks.
Tonight, we reached an agreement with the Department of War to deploy our models in their classified network.
In all of our interactions, the DoW displayed a deep respect for safety and a desire to partner to achieve the best possible outcome.
AI safety and wide distribution of…
— Sam Altman (@sama) February 28, 2026
OpenAI also called on the Department of Defense to offer the same terms to all AI companies, believing all businesses should be willing to accept them. Altman also emphasized that OpenAI strongly hopes the situation will de-escalate, shifting from legal and government action to rational negotiation.
Anthropic Clashes with Pentagon Over ‘Unprecedented’ Move, Vows Legal Action
In stark contrast to OpenAI’s conciliatory stance, Anthropic’s tensions with the Pentagon continue to escalate. According toReportAnthropic said it has not received any direct communication from the “Department of War” or the White House regarding the progress of negotiations, and explicitly stated it will challenge in court any determination that classifies it as a supply chain risk. US President Trump said Friday that he has directed US government agencies to stop using Anthropic’s technology. The Pentagon subsequently announced that it had designated Anthropic as a supply chain risk company. Anthropic criticized sharply in a statement:
「將 Anthropic 列為供應鏈風險將是一個前所未有的舉動，此舉歷來只針對美國的對手，此前從未公開適用於任何美國公司。我們對這一事態的發展深感痛心。」
According to sources, the Pentagon wants to use Anthropic’s Claude chatbot without restrictions within the bounds of the law. However, Anthropic has consistently maintained that Claude must not be used for mass surveillance of Americans or for operating fully autonomous weapons. The Pentagon subsequently issued an ultimatum to Anthropic: relax the restrictions by Friday or lose the contract. But Anthropic chose to refuse.
Over 300 Google, OpenAI Employees Sign Rare Joint Letter Supporting Anthropic
Just as the compliance deadline (Friday afternoon) given to Anthropic by the Pentagon was about to expire, the situation took a dramatic turn. Over 300 Google employees and more than 60 OpenAI employeesjointly signedOpen LetterThey urged their respective company leadership to support Anthropic and reject such unilateral usage demands. The open letter specifically called on Google and OpenAI executives to uphold Anthropic’s red lines regarding mass surveillance and fully autonomous weapons. The letter stated:
「他們試圖用『對方會妥協』的恐懼來分化每家公司。這種策略只有在我們互不了解彼此立場時才會奏效。」
「我們希望領導層能放下分歧，團結一致，繼續拒絕『戰爭部』當前的需求。」
The signatories of this open letter, spontaneously signed by employees of AI giants, emphasize their support for Anthropic’s stance against using AI for domestic mass surveillance and autonomous weapons. They hope to encourage their respective employers to “set aside differences and unite as one” and uphold the ethical boundaries outlined by Anthropic. So far, executives at both companies have not formally responded to the letter.
Perspective: The “Militarization” Trend in the AI Industry and Ethical Red Lines
This series of events reveals a profound dilemma facing the AI industry: when national security and commercial interests intertwine, how can tech companies balance “cooperating with the government” against “sticking to their values”? From OpenAI’s perspective, choosing to reach an agreement with the Department of Defense and writing security principles into the contract may be a “pragmatic” survival strategy. By ensuring two core red lines—banning mass surveillance and requiring human responsibility for weapons use—OpenAI maintains cooperation while preserving a moral bottom line.
However, Anthropic’s hardline stance represents an alternative choice—even in the face of being labeled a supply chain risk and potentially losing government contracts, it remains resolutely unwilling to compromise. Anthropic’s position is clear: AI must not be used to harm innocent people or serve as a tool for authoritarian surveillance.
As for the joint letter signed by Google and OpenAI employees, it demonstrates that concerns about the “militarization of AI” are brewing within the tech industry. By choosing to publicly side with Anthropic, the employees are not only showing solidarity with their peers, but also exercising a form of “internal oversight” of their own companies.
Key points to watch:
- Will Anthropic really file a lawsuit against the Department of Defense’s decision?
- Will OpenAI’s collaboration with the Department of Defense set a precedent that other AI companies will follow?
- How will Google and OpenAI executives respond to their employees’ public appeal?
This debate about AI ethics and defense is really just beginning. Do you think tech giants’ AI should be banned from defense projects? Honestly, I’m quite cautious about this. If AI really does “wake up” one day and becomes something like Skynet from Terminator, and it can control these unmanned weapons of mass destruction, that would really be terrifying.
Source: KOCPC Chinese