Anthropic AI Faces Pete Hegseth Ultimatum as Pentagon Demands Military Access to Claude

An unprecedented showdown between the Trump administration and one of the world's leading AI companies erupted this week. Anthropic, the maker of the Claude AI model, received a Friday deadline from Defense Secretary Pete Hegseth to strip away its safety guardrails for military use — or face the consequences. The clash has sent shockwaves across the AI industry and Wall Street alike.

Pete Hegseth Gives Anthropic a Friday Deadline

Defense Secretary Pete Hegseth gave Anthropic CEO Dario Amodei a Friday deadline to comply with demands to peel back safeguards on its AI model or risk losing a $200 million Pentagon contract. He also threatened to place Anthropic on what could amount to a government blacklist. At issue are the guardrails Anthropic placed on Claude — the Pentagon wants the company to lift restrictions for "all lawful use."

Anthropic has two concerns it is not willing to drop: AI-controlled autonomous weapons and mass domestic surveillance of American citizens. The company believes AI is not reliable enough to operate weapons, and argues there are no laws yet covering how AI could be used in mass surveillance programs.

The Pentagon Meeting: Cordial but Tense

During the Tuesday morning meeting, a senior Defense official described the encounter as "not warm and fuzzy at all," while another source said it remained "cordial" with no raised voices. Hegseth praised Claude and told Amodei he would not allow any company to dictate the terms under which the Pentagon makes operational decisions or object to individual use cases.

Hegseth specifically raised the Pentagon's claim that Anthropic had objected to the use of Claude during the Maduro raid via partner Palantir. Amodei denied that Anthropic raised any such concerns, stating the company's red lines have never prevented the Pentagon from doing its work or posed an issue for anyone in the field.

Anthropic Refuses to Budge on AI Safety

Anthropic has no plans to ease its usage restrictions, and made clear it is prepared to walk away from negotiations if its concerns about autonomous weapons and mass surveillance of Americans are not addressed. The company was the first frontier AI lab with classified Department of Defense access and the only one currently deployed on the most sensitive military networks.

The Department of Defense has no backup option currently in play, creating a single-vendor dependency that weakens the Pentagon's hand. One defense consultant noted the DOD cannot fix that vulnerability overnight, adding that the situation reflects a serious institutional miscalculation.

Pete Hegseth's Leverage: Defense Production Act and Supply Chain Risk

If Anthropic fails to comply by the Friday evening deadline, Hegseth threatened to label the company a "supply chain risk" — a designation typically reserved for foreign adversaries — which would require DoD vendors and contractors to certify they do not use Anthropic's models anywhere in their workflows. He also threatened to invoke the Defense Production Act, which gives the president authority to compel private companies to accept government contracts in the interest of national security.

Elon Musk's xAI recently agreed to allow its models to be deployed in classified Pentagon settings, making Grok the second AI model cleared for such use. The White House is also speeding up conversations with OpenAI and Google to potentially move their models into classified systems as alternatives to Claude.

Anthropic's $380 Billion Valuation and Enterprise Expansion

Even as the standoff with Pete Hegseth intensified, Anthropic made headlines this week for other major moves. Anthropic raised $30 billion in Series G funding led by GIC and Coatue, valuing the company at $380 billion post-money, with the investment aimed at fueling frontier research, product development, and infrastructure expansion.

On Tuesday, Anthropic launched new updates to its Claude Cowork AI agent integrating with enterprise apps including Salesforce-owned Slack, Intuit, Docusign, LegalZoom, FactSet, and Google's Gmail. Software stocks that had sold off sharply in recent weeks largely rebounded, with analysts at Wedbush Securities calling the selloff "overblown."

Anthropic Loosens Safety Policy Amid AI Race

Separately, Anthropic updated its Responsible Scaling Policy, originally published in 2023, saying it would no longer delay potentially dangerous AI development if the company believes it lacks a significant lead over a competitor. The change reflects growing competitive pressure from rivals including OpenAI and Google.

The Friday deadline from Pete Hegseth now looms as a defining moment for both the future of Anthropic's government contracts and the broader question of how much control Washington can exercise over the AI industry's safety standards.