The escalating dispute between the Pentagon and artificial intelligence firm Anthropic is not just a contract fight. It exposes the deeper struggle shaping U.S. national security. In my forthcoming book, The New AI Cold War, I argue that artificial intelligence is becoming indispensable to national power. The friction between Silicon Valley guardrails and military necessity shows that transformation is already underway.
At issue is how the U.S. military used Anthropic’s AI model, Claude, during the January 2026 operation that captured Venezuelan leader Nicolás Maduro. Reporting from The Hill confirms the Pentagon is reviewing its relationship with Anthropic after reports that Claude was used through Palantir during the mission. PCMag likewise reported tensions over how broadly the military may deploy the system.
The disagreement centers on guardrails. Anthropic maintains restrictions against uses such as fully autonomous weapons and mass domestic surveillance. The Pentagon insists that partners must allow use for “all lawful purposes.” As Pentagon spokesman Sean Parnell put it, America requires partners willing to help “our warfighters win in any fight.”
That’s the strategic reality. AI is no longer confined to chatbots or productivity tools. It is embedded in daily military operations. Claude’s presence on classified networks demonstrates how deeply frontier AI models are now woven into intelligence workflows.
These systems can process satellite imagery, intercepted communications, sustainment data, and financial tracking at speeds no human staff can match. They identify patterns, flag anomalies, and compress decision timelines from days to minutes. In high-risk missions, that time compression can determine success or failure.
But the power of AI comes from data fusion — integrating vast, disparate streams into one operational picture. And that same power creates temptation. A system designed to track foreign weapons networks could, over time, be granted access to domestic communications metadata or biometric databases. Mission creep rarely arrives with fanfare. It expands quietly, one new data permission at a time.
That’s what makes this debate serious.
Most Americans will never see a classified AI dashboard. But they will live under the legal framework governing it. The guardrails set for military AI today will influence how algorithmic power is used across government tomorrow. This is not merely a defense contracting dispute. It is a constitutional question.
Meanwhile, America’s adversaries are moving forward aggressively.
China’s military modernization strategy emphasizes “intelligentized warfare” (智能化战争), integrating AI into command systems, autonomous platforms, and decision-support tools. The People’s Liberation Army has incorporated AI across its C4ISR architecture and publicly demonstrated autonomous strike concepts at the Zhuhai Air Show. Russia has invested heavily in AI-enabled cyber operations, electronic warfare, and battlefield reconnaissance. Beijing and Moscow are not designing these systems with constitutional limits in mind.
Here is the hard truth: if the United States hesitates to harness AI for defense, competitors will not.
There’s a second threat here.
Anthropic CEO Dario Amodei recently warned that democracies must draw firm lines around AI misuse. He is right. The same technology that accelerates targeting abroad can enable intrusive surveillance at home.
What would that look like in practice?
Imagine an AI system trained to analyze millions of financial transactions, travel records, and social media posts simultaneously. The tool that tracks a foreign terror network overseas could, if redirected inward, flag Americans for “suspicious behavior” based on algorithmic correlations rather than probable cause. It could map relationships between citizens, predict protest activity, or generate automated threat assessments without a warrant. The technology itself is neutral. Its application is not.
The danger also applies to battlefield autonomy. The Department of Defense’s Directive 3000.09 requires appropriate human judgment over lethal force. Remove that safeguard, and the implications change dramatically. AI-enabled drone swarms could identify and engage targets based solely on pattern recognition — heat signatures, movement profiles, facial matches — without a human confirming the final decision. Delegating greater authority to machines may increase speed. It also increases the risk of error and unintended escalation.
The Pentagon’s insistence on flexibility reflects strategic necessity. Anthropic’s caution reflects constitutional concern. Both deserve thoughtful consideration.
We must be strong — and disciplined.
Congress must establish clear statutory boundaries for military AI use. Start with three common-sense principles.
First, human accountability must remain central in lethal decisions. AI may assist, but commanders remain responsible under the law of armed conflict.
Second, domestic AI-enabled surveillance must be subject to judicial oversight. The Fourth Amendment does not disappear because an algorithm is involved.
Third, defense AI systems should incorporate transparency and auditability wherever feasible, ensuring decisions can be reviewed after action.
America’s strength has always come from pairing innovation with constitutional restraint. That is what distinguishes us from authoritarian regimes building AI systems for surveillance and coercion.
AI is already reshaping national security. The question is whether Washington can keep up.
If the United States fails to lead in capability, we risk ceding strategic advantage to Beijing and Moscow. If we fail to lead in governance, we risk eroding the liberties that define us.
We must do both.
In the emerging AI Cold War, technological dominance and moral clarity must advance together. Security without liberty is not the American model. But liberty without security is unsustainable.
Artificial intelligence is now a central instrument of national power. The challenge before us is to harness it wisely — strong enough to defend us — and wise enough not to turn that power inward.
Notice: This column is printed with permission. Opinion pieces published by AFN.net are the sole responsibility of the article's author(s), or of the person(s) or organization(s) quoted therein, and do not necessarily represent those of the staff or management of, or advertisers who support the American Family News Network, AFN.net, our parent organization or its other affiliates.
