Tension at the Pentagon is rising as the role of artificial intelligence in U.S. national security is being defined. Defense Secretary Pete Hegseth has summoned Dario Amodei, CEO of Anthropic, for a key meeting at the Department of Defense next Tuesday. Close sources indicate that this is not just a simple meeting, but a decisive negotiation regarding the military use of Claude, Anthropic's advanced AI model. Currently, Claude is the only AI system integrated into classified operations of the U.S. military, although this exclusivity may soon change. Claude's ability to analyze sensitive information makes it a strategic tool, but also a source of conflict. Military leaders value its power but are frustrated by Anthropic's refusal to remove certain ethical and operational safeguards that restrict its use in combat. Both parties enter the negotiation with opposing views on technological autonomy. Anthropic insists that the dialogue is progressing constructively, while defense officials believe the process is stalled and close to failure. The company has shown a willingness to relax some restrictions but maintains two non-negotiable limits: Claude cannot be used for mass surveillance of U.S. citizens or for developing autonomous weapons that operate without human oversight. Amodei reiterated this in a recent statement. The Pentagon, for its part, believes these limitations hinder operability and demands full access to AI models for any legal use, without requiring Anthropic's approval for each application. Pressure has escalated to the point where the Pentagon threatens to declare Anthropic a risk to the supply chain, which would lead to contract cancellations and require other companies to certify that they do not use Anthropic technology. This confrontation also reflects a cultural clash between the tech companies' security vision and military logic. Amodei has publicly warned about the risks of uncontrolled AI, advocating for a responsible approach, while the Pentagon views these precautions as an ideological obstacle to national defense needs. A recent incident, where Claude was used in an operation to capture Nicolás Maduro, intensified the debate over the ethical and strategic limits of AI. Anthropic's stance is supported by hundreds of employees from OpenAI and Google, who have signed an open letter backing Amodei's decision. Now, the Pentagon's response will be crucial in a conflict that tests the balance between technological ethics and national security.