WASHINGTON – A high-stakes standoff between the Trump administration and Anthropic, a leading artificial intelligence company, is nearing a critical juncture. The Pentagon has issued an ultimatum: bend its ethical policies regarding the use of its AI model, Claude, or risk losing a lucrative $200 million contract. The dispute centers on the military’s desire to utilize Claude without limitations, a proposition Anthropic CEO Dario Amodei has firmly rejected, stating the company “cannot in good conscience accede” to requests that would allow the technology to be used for lethal operations. This clash raises fundamental questions about the role of tech companies in shaping the future of warfare and whether they can, or should, dictate the terms of their technology’s deployment.
The core of the disagreement lies in Anthropic’s commitment to AI safety and its refusal to compromise those principles, even under pressure from the Department of Defense. Claude, currently the top-performing AI model for sensitive intelligence activities, is uniquely positioned within the military’s most secure systems. The Pentagon’s demand for unrestricted access has sparked a broader debate about the ethical boundaries of artificial intelligence and the potential for autonomous weapons systems. The situation highlights a growing tension between national security interests and the moral responsibilities of AI developers, a dynamic that could redefine the relationship between government and the tech sector. This case involving Anthropic and the U.S. Government is prompting a wider discussion about whether Big Tech companies can fundamentally rewrite the social contract.
A Latest Power Dynamic: Tech Companies as Stakeholders
According to Father Paolo Benanti, president of the Ethical Committee for the Adoption of Artificial Intelligence in the World of Work, established by the Italian government, the Anthropic case underscores a significant shift in power dynamics. “The case of Anthropic highlights the new dimension and importance of technology players, which boast numbers that they didn’t have before,” Benanti explained. He noted that nine of the world’s top ten companies are now technology firms. “It too raises the issue of differing perceptions among stakeholders in the American public debate.” Benanti frames the situation as a question of whether the government should be considered simply another client, and if so, whether its requested uses of Anthropic’s technology violate the existing service agreement.
The ethical considerations extend beyond simply whether military use of AI is inherently wrong. Benanti argues that the core issue is one of democratic balance. “Does the government have the right to use it as it wants, simply because it is the government?” he asks, acknowledging that there are no straightforward answers. He points to a 2025 instance where Microsoft denied services to the Israeli government due to concerns about human rights violations and adherence to its terms of service. This precedent demonstrates that even governments are not exempt from a technology partner’s contractual obligations.
Can the U.S. Government Override Contractual Agreements?
The current situation with Anthropic represents a step further, Benanti explains, because it involves the U.S. Government itself being dependent on an American service provider. The question then becomes: can the “national interest” supersede Anthropic’s contractual exception? The U.S. Government has previously sought access to data held by tech companies, as evidenced by a case involving Microsoft’s cloud services and a narcotics investigation. That dispute ultimately led to the passage of the Cloud Act, clarifying the circumstances under which law enforcement can access cloud data held by American companies. This ongoing series of cases is gradually defining the relationship between technology partners and governmental institutions.
Benanti suggests that the broader implication of these events is whether technology partners are poised to “write the social contract.” “So far, this has never been the case, it would be a nice novelty,” he said. “This is the real question. We are facing a new subject: it doesn’t exist in Montesquieu’s theory of the balance of powers.” Whether this shift will occur remains to be seen, but Benanti emphasizes the importance of observing these unfolding events. If the social contract is increasingly shaped by technology companies, with the state relegated to the role of a client, it raises profound democratic concerns.
AI and the Future of Work: Lessons from Italy
The debate surrounding AI extends beyond military applications. In Italy, the use of artificial intelligence in the labor market has sparked controversy, leading to the government’s intervention in cases involving Deliveroo and Glovo. These companies were placed under administration by the judiciary due to alleged violations of workers’ rights. “An unmanaged algorithm is no less ruthless than an unmanaged employer, or an unmanaged labor market,” Benanti stated. He advocates for the regulation of digital tools to ensure they adhere to the same standards that protect workers, emphasizing that Italy still maintains sovereignty over its labor laws.
The situation with Anthropic, and the broader implications for the role of technology companies in society, are still developing. The Pentagon’s deadline for Anthropic is Friday, February 27, 2026, and the outcome will likely set a precedent for future interactions between the government and AI developers. The next step will be to spot whether Anthropic maintains its stance, potentially jeopardizing the $200 million contract, or if a compromise can be reached that addresses both national security concerns and ethical considerations. This case serves as a crucial test of whether the principles of responsible AI development can withstand the pressures of geopolitical interests.
What are your thoughts on the ethical implications of AI in national security? Share your perspective in the comments below and join the conversation.
Related reading
