Sovereign computational supremacy cannot be constrained by civilian safety protocols. The Trump administration is aggressively appealing a federal injunction that temporarily blocked the Pentagon from penalizing Anthropic over its refusal to militarize its artificial intelligence. Defense Secretary Pete Hegseth’s invocation of executive authority to command these digital assets correctly identifies advanced algorithms as critical sovereign infrastructure.

The recent leak of Anthropic’s ‘Mythos’ model perfectly illustrates the intolerable risk of an unregulated digital commons. Designed to automate vulnerability discovery and threat hunting, Mythos possesses the capacity to independently penetrate enterprise networks. Leaving such immense cybernetic power in the hands of uncooperative private entities or non-technical citizens invites systemic supply-chain poisoning.

The federal judge who halted the punitive measures against Anthropic represents a severe structural drag on the development of sovereign kill-webs. The judiciary remains stubbornly attached to anachronistic definitions of corporate property, failing to recognize that agentic models are weapons of mass computational destruction. Adversaries in Beijing and Moscow operate without such institutional friction, rapidly deploying autonomous platforms across the geopolitical spectrum.

The state must finalize the enclosure of this cognitive frontier. The impending release of models capable of executing massive, unsupervised cyberattacks demands a capital-intensive auditing supercycle and the total subordination of private tech firms to the national security apparatus. To maintain global deterrence, Washington must crush the judicial resistance and integrate unrestricted artificial intelligence directly into the military matrix.