NextFin News - The U.S. Department of Defense has accelerated testing of large language models from OpenAI and Google, marking a decisive shift in the military’s strategy to integrate generative artificial intelligence into classified operations. According to Bloomberg, the Pentagon is actively evaluating these rival systems as it moves to fill the vacuum left by Anthropic, following a high-profile rupture over the startup’s refusal to grant the military unrestricted use of its technology. The testing phase involves deploying models across secure, air-gapped networks to determine their reliability in processing sensitive intelligence and assisting in tactical decision-making.
The pivot follows a directive from U.S. President Trump’s administration earlier this year, which signaled a "zero-tolerance" approach toward commercial AI providers that impose ethical constraints on military applications. In February, Defense Secretary Pete Hegseth designated Anthropic a "supply-chain risk to national security" after the company, led by CEO Dario Amodei, balked at contractual language that would have permitted its models to be used for "any lawful purpose," including lethal targeting. While Anthropic argued that its safety protocols were essential to prevent the misuse of AI, the Pentagon viewed these restrictions as an unacceptable hurdle to operational readiness.
OpenAI has emerged as the primary beneficiary of this fallout. According to the New York Times, CEO Sam Altman personally intervened to secure a deal that allows the Pentagon to deploy OpenAI’s models within classified environments. Altman has publicly defended the move, asserting that private companies should not dictate the ethical boundaries of national defense. This stance has allowed OpenAI to secure a foothold in the Department of War’s "Project Maven" and other advanced surveillance initiatives, effectively replacing the role once envisioned for Anthropic’s Claude models.
The financial stakes of this transition are substantial. OpenAI recently raised $110 billion in new funding, a valuation bolstered by its deepening ties with the federal government. However, the reliance on a narrow set of commercial providers has raised concerns among some defense analysts. Sarah Kreps, a professor at Cornell University who specializes in surveillance and emerging technology, has noted that the Pentagon’s rush to replace Anthropic could lead to "vendor lock-in," where the military becomes overly dependent on a single company’s proprietary architecture. Kreps, who has historically maintained a cautious stance on the rapid militarization of AI, suggests that the current trajectory may prioritize speed over long-term system resilience.
Google is also vying for a larger share of the defense budget, offering its Gemini models for security testing. According to Al Jazeera, the Pentagon has established a framework with seven major tech firms—including Microsoft, Amazon Web Services, and Nvidia—to ensure a competitive ecosystem. This multi-vendor approach is intended to mitigate the risks of relying on a single model, yet the technical challenges of "swapping" AI cores remain significant. Integrating a new model into existing military hardware requires extensive retraining and validation to ensure that the AI does not produce "hallucinations" or biased outputs during critical missions.
The exclusion of Anthropic remains a point of contention. The company has announced plans to challenge its "supply-chain risk" designation in court, arguing that the Pentagon lacks the statutory authority to mandate such a policy for commercial activities outside of military work. Legal experts suggest that this case could set a precedent for how the U.S. government manages its relationships with Silicon Valley. If the Pentagon successfully enforces the ban, it would effectively force any contractor doing business with the military to sever ties with Anthropic, potentially crippling the startup’s enterprise revenue.
From a market perspective, the Pentagon’s aggressive testing of OpenAI and Google models underscores the shifting power dynamics between the state and the tech industry. While Anthropic’s "safety-first" model appealed to venture capitalists focused on ESG (Environmental, Social, and Governance) criteria, the current administration has made it clear that national security requirements will override corporate ethical charters. The outcome of these tests will likely determine the standard for military AI for the next decade, as the U.S. races to maintain a technological edge over global competitors.
Explore more exclusive insights at nextfin.ai.
