OpenAI and Anthropic: A Shifting Landscape in Pentagon AI Contracts
The US Department of Defense (DoD) has undergone a significant shift in its AI contracting strategy, with OpenAI stepping in to replace Anthropic as a key supplier of AI technology. This development follows a week of public disagreements between the Trump administration and leading AI companies, highlighting the complex challenges of integrating powerful new technologies with national security concerns.
The Dispute with Anthropic
The core of the dispute centered on Anthropic’s insistence that its AI models not be used for “mass surveillance” or in the development of “fully autonomous weapons.” Defense Secretary Pete Hegseth dismissed these provisions as “woke,” signaling a fundamental disagreement over the acceptable uses of AI technology. On February 27, 2026, President Donald Trump ordered federal agencies to cease using Anthropic’s models and the DoD designated the company as a “supply chain risk,” effectively barring it from future government contracts and impacting its ability to function with government contractors and suppliers .
OpenAI Steps In
Within hours of the Trump administration’s actions against Anthropic, OpenAI, led by CEO Sam Altman, announced a deal to provide its AI tools for apply in classified military systems. Altman stated that the agreement included similar guardrails to those Anthropic had requested, specifically prohibiting domestic mass surveillance and ensuring human responsibility for the use of force, including autonomous weapon systems . Altman emphasized the importance of technical safeguards and proposed that the DoD extend these terms to all AI companies .
A Market Dynamic and Branding Considerations
Despite the public posturing, some analysts suggest this outcome may be beneficial for both Anthropic and the Pentagon. AI models are becoming increasingly commoditized, with performance differences between leading providers like Anthropic, OpenAI, and Google being relatively minor. In this market, branding and perceived trustworthiness are crucial. Anthropic has positioned itself as the morally responsible AI provider, a stance that may hold value with consumers and enterprise clients. OpenAI, by taking Anthropic’s place in government contracting, risks politicizing its brand and potentially alienating customers who support Anthropic’s ethical stance .
The Pentagon’s Options and the Rise of Open-Weight Models
The Pentagon has alternative options beyond relying on major tech companies. The department has already deployed dozens of open-weight models – AI models with publicly available parameters – which are often licensed permissively for government use. This demonstrates the DoD’s ability to function even without the cooperation of leading AI developers.
A Broader Context: AI and National Security
The situation highlights the broader debate surrounding the use of AI in military applications. The development of autonomous weapons systems is ongoing, evolving from primitive traps to sophisticated, radar-guided systems like the 1980s-era Phalanx CIWS. AI will inevitably be integrated into military technologies, raising fundamental questions about ethical considerations and democratic control .
The Need for Democratic Oversight
the dispute underscores the importance of robust democratic structures and legal frameworks to govern the development and deployment of AI. If the DoD seeks to utilize AI for purposes that the public finds unacceptable, such as mass surveillance or autonomous warfare, new legal restrictions are needed. Strengthening legal protections around government procurement is also crucial to prevent the government from dictating unsafe applications of AI technology.
The Pentagon should prioritize maximizing its warfighting capabilities within the bounds of the law, while private companies like Anthropic should focus on building consumer and buyer confidence. However, neither should be assumed to be acting solely in the public’s interest.