Hegseth Pressures Anthropic to Lift Military Restrictions on AI Models

The reported warning from Defense Secretary Pete Hegseth to Anthropic marks a significant escalation in the Department of Defense's (DoD) efforts to harness commercial artificial intelligence for national security. According to sources familiar with the matter, the Pentagon is no longer content with the restrictive 'Acceptable Use' policies that leading AI labs have traditionally applied to their models. Hegseth’s demand that Anthropic allow the military to use its technology 'as it sees fit' suggests a pivot toward a more coercive relationship between the federal government and the Silicon Valley firms currently leading the generative AI revolution.

Anthropic, founded by former OpenAI executives with a core mission of 'AI safety,' has long been a proponent of Constitutional AI—a method of training models to follow a specific set of ethical principles. Historically, these principles have included prohibitions against using AI for lethal autonomous weapons, surveillance that violates civil liberties, or direct kinetic military operations. However, as the geopolitical race for AI supremacy with China intensifies, the DoD increasingly views these self-imposed ethical restrictions as strategic liabilities. The Pentagon's argument is rooted in the belief that if American developers do not provide the military with unrestricted access to the world’s most advanced large language models (LLMs), the United States will inevitably lose its technological edge to adversaries who operate without such moral constraints.

“

The reported warning from Defense Secretary Pete Hegseth to Anthropic marks a significant escalation in the Department of Defense's (DoD) efforts to harness commercial artificial intelligence for national security.

This development places Anthropic in a precarious position. The company has raised billions of dollars from investors like Amazon and Google, positioning itself as the 'safe' alternative to more aggressive competitors. If Anthropic bows to Hegseth’s pressure, it risks a backlash from its workforce—many of whom joined the company specifically because of its safety-first ethos—and could face a reputational crisis among enterprise clients who value its commitment to ethical AI. Conversely, defying the Pentagon could lead to more heavy-handed regulatory actions. Analysts suggest the administration could potentially invoke the Defense Production Act (DPA) to compel cooperation, or use the threat of excluding the company from massive federal procurement contracts to force a policy shift.

The broader industry context is equally fraught. While OpenAI recently softened its stance on military collaboration—partnering with the DoD on cybersecurity initiatives—it still maintains a line against 'weapons development.' Hegseth’s specific targeting of Anthropic suggests the Pentagon is looking to break the strongest link in the chain of AI safety advocates. If Anthropic yields, it sets a precedent that commercial AI safety policies are secondary to national security mandates. This would effectively transform the 'AI safety' movement from a private governance model into a state-directed regulatory framework where the definition of 'safe' is determined by the requirements of the battlefield.

Looking forward, the industry should watch for a formal response from Anthropic’s leadership, including CEO Dario Amodei. The company may attempt to negotiate a middle ground—perhaps creating a 'defense-specific' version of its Claude model that is air-gapped or fine-tuned for tactical analysis while maintaining restrictions on lethal targeting. However, Hegseth’s 'as it sees fit' phrasing leaves little room for such nuance. This standoff is likely the opening salvo in a long-term struggle over who controls the 'brain' of modern warfare: the engineers who build the models or the commanders who deploy them. The outcome will define the operational limits of AI in the 21st century and could determine whether the next generation of defense technology is built on a foundation of commercial ethics or military necessity.