Pentagon Issues Final Ultimatum to Anthropic: Military AI Access or Defense Production Act

The U.S. Department of Defense has set a strict deadline of 5:01 PM ET today, Friday, February 27, for AI developer Anthropic to grant the military unrestricted access to its technology for "all lawful purposes." Defense Secretary Pete Hegseth has warned that the government is prepared to invoke the Cold War-era Defense Production Act or designate the firm as a "supply chain risk" if it continues to enforce ethical restrictions on military applications. This unprecedented standoff represents a critical turning point in the relationship between Silicon Valley's private AI sector and the U.S. national security apparatus.

Hegseth's Ultimatum and the 5:01 PM Deadline

In a tense meeting at the Pentagon on Tuesday, Defense Secretary Pete Hegseth delivered a blunt message to Anthropic CEO Dario Amodei: the military will no longer accept commercial terms that limit its operational capabilities. Sources familiar with the discussion report that Hegseth gave the company until the close of business today to remove its specific "red lines"—contractual clauses that currently prohibit the use of its Claude AI models for mass surveillance and fully autonomous weapons systems.

The ultimatum marks a significant escalation in the Pentagon's strategy to integrate cutting-edge artificial intelligence into defense operations. "We are done asking nicely," a senior defense official reportedly stated. "National security isn't a suggestion." If Anthropic refuses to comply, the Department of Defense has threatened to invoke the Defense Production Act (DPA). Originally enacted in 1950 during the Korean War, the DPA gives the President broad authority to compel private companies to prioritize government contracts and potentially force the modification of products to meet national defense needs.

The Threat of "Supply Chain Risk" Designation

Beyond the DPA, the Pentagon is wielding a potentially more damaging economic weapon: the threat of designating Anthropic a "supply chain risk." This classification, typically reserved for foreign adversaries like Huawei or ZTE, would effectively blacklist the company from the entire federal marketplace. The impact would ripple far beyond direct military contracts; prime defense contractors like Boeing and Lockheed Martin would likely be forced to strip Anthropic's technology from their own systems to maintain their eligibility for government work.

Demonstrating that this is not an idle threat, the Pentagon has already taken preliminary steps this week by formally requesting that major defense contractors assess and report their current exposure to Anthropic's technology. This move signals to the market that the military is preparing for a future without Anthropic if the company does not capitulate.

The "Maduro Raid" Catalyst and Operational Disputes

The conflict has been fueled by a specific, classified event that brought these theoretical disagreements into sharp relief. Reports confirm that in January 2026, U.S. special operations forces utilized Anthropic's Claude AI model during a high-stakes mission to capture Venezuelan President Nicolás Maduro. The AI was accessed through a partnership with data analytics firm Palantir, reportedly to analyze real-time intelligence.

Following the operation, Anthropic raised concerns that this usage may have violated its Acceptable Use Policy, which strictly forbids the deployment of its models in scenarios involving lethal force or autonomous targeting. This retroactive objection infuriated Pentagon leadership, who view such restrictions as an intolerable operational liability. For Defense Secretary Hegseth, the incident underscored the danger of relying on a vendor that claims the right to veto specific military missions based on its own corporate ethics.

Anthropic's "Responsible Scaling" vs. Military Reality

Caught in the crosshairs is Anthropic's core identity as a safety-first AI lab. On Tuesday—the same day as the heated Pentagon meeting—the company released Version 3.0 of its "Responsible Scaling Policy" (RSP). While the update softened some previous commitments to unilaterally pause development, citing intense competitive pressure from rivals who are "blazing ahead," the company has publicly maintained its refusal to cross its ethical red lines regarding military use.

CEO Dario Amodei has argued that these restrictions are not about hindering national defense but about preventing the creation of uncontrollable automated warfare systems. However, the ground is shifting beneath him. Competitor xAI, led by Elon Musk, has already agreed to the Pentagon's "all lawful purposes" standard for its Grok model, effectively isolating Anthropic as the lone dissenter among major AI labs.

What Happens Next?

As the 5:01 PM deadline approaches, the industry holds its breath. Capitulation would mean the end of Anthropic's unique "constitutional AI" stance, potentially alienating its safety-focused employee base. Refusal, however, could lead to a legal and economic siege that few companies could withstand. If the Pentagon invokes the Defense Production Act to seize control of a software model's parameters, it would set a historic legal precedent, effectively nationalizing the ethical guardrails of private AI technology for the sake of 2026 national security objectives.