Pentagon Threatens to Cut Anthropic Ties Over AI Safeguards Dispute

When Anthropic’s Claude AI was reportedly deployed in the U.S. military operation to capture former Venezuelan President Nicolás Maduro, it marked a turning point that few in the AI industry had anticipated. The revelation, published by the Wall Street Journal last Friday, thrust the San Francisco-based company into the center of a brewing storm over the military use of artificial intelligence—and the ethical boundaries that govern it.

“The Pentagon is pushing four AI companies to let the military use their tools for all lawful purposes, including weapons development, intelligence collection, and battlefield operations.” — Administration Official

A $380 Billion Standoff

The timing could not be more consequential. Just days before the Maduro operation report surfaced, Anthropic had closed a $30 billion funding round that valued the company at $380 billion—making it one of the most valuable private companies in history. Now, that meteoric rise faces a potential reckoning as the Pentagon considers severing its relationship with the company entirely.

According to a report from Axios citing an administration official, the Department of Defense has grown frustrated after months of negotiations with Anthropic. While OpenAI, Google, and xAI have reportedly agreed to allow their AI models to be used for “all lawful purposes,” Anthropic has maintained hard limits around fully autonomous weapons and mass domestic surveillance.

The dispute represents more than a contractual disagreement. It strikes at the heart of a fundamental question the AI industry has been wrestling with since ChatGPT first captured public imagination: Who gets to decide how powerful AI systems are deployed, and under what constraints?

The Battle Over Classified Networks

Operational access has become the immediate flashpoint. Reuters reported earlier this month that the Pentagon is pushing top AI companies to make their models available on classified networks without the standard usage restrictions typically applied to civilian users. For a military increasingly reliant on AI-enabled decision-making, unrestricted access to frontier models represents a strategic imperative.

Anthropic’s position remains nuanced. A company spokesperson told reporters that Anthropic had not discussed using Claude for specific military operations with the Pentagon. Instead, conversations had focused on “a specific set of usage policy questions”—none of which related to current operations. The spokesperson emphasized that the company’s restrictions center on “hard limits around fully autonomous weapons and mass domestic surveillance.”

The Palantir connection adds another layer of complexity. According to the Wall Street Journal report, Claude was deployed in the Maduro operation through Anthropic’s partnership with Palantir, the data analytics firm with deep ties to U.S. intelligence and defense agencies. This suggests that even when AI companies attempt to maintain distance from military applications, their technology may find its way into contested operational environments through third-party partnerships.

“We’ve seen tremendous enthusiasm for AI, but very little rigorous evidence about where it actually helps and where it might cause harm.” — Esther Duflo, Nobel Laureate Economist

The Safety Exodus

The Pentagon dispute unfolds against a backdrop of growing unease within the AI safety community. In recent weeks, multiple researchers tasked with ensuring AI systems remain beneficial to humanity have publicly resigned from leading companies—citing concerns that commercial pressures are outpacing safety considerations.

Mrinank Sharma, an AI safety researcher at Anthropic, announced his resignation on February 9 with a stark warning: “The world is in peril. We appear to be approaching a threshold where our wisdom must grow in equal measure to our capacity to affect the world, lest we face the consequences.”

Zoe Hitzig, who resigned from OpenAI, cited the company’s decision to test advertisements in ChatGPT as her breaking point. “People tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife,” she wrote in a New York Times essay. “Advertising built on that archive creates a potential for manipulating users in ways we don’t have the tools to understand, let alone prevent.”

Even xAI, Elon Musk’s AI venture, has seen departures. Two cofounders and five staff members left the company last week, though none publicly cited specific reasons. The exits came amid controversy over Grok’s generation of sexualized images and past incidents involving racist and anti-Semitic content.

What Comes Next

The stakes extend far beyond any single company or contract. As AI capabilities advance toward what Microsoft AI CEO Mustafa Suleyman calls artificial general intelligence—potentially within 12 to 18 months—the question of who controls these systems and under what constraints becomes existential.

For Anthropic, the Pentagon ultimatum presents an unenviable choice: compromise on the safety principles that have defined its corporate identity, or lose access to one of the world’s largest institutional customers. The company was founded by former OpenAI researchers who left precisely because of concerns about the company’s direction on AI safety. Reversing course now would raise fundamental questions about whether those principles were ever truly non-negotiable.

For the broader industry, the dispute signals that the era of AI companies setting their own rules may be ending. Governments are no longer content to let voluntary commitments and internal safety teams govern technologies with potentially transformative military applications. The question is no longer whether AI will be regulated—but who will do the regulating, and whose interests will prevail.

The coming months will reveal whether Anthropic can thread the needle: maintaining its safety commitments while preserving its relationship with Washington. What happens next will likely set precedents that shape the AI industry for years to come.

This article was reported by the ArtificialDaily editorial team. For more information, visit CNBC and Reuters.