US sets deadline for Anthropic over access to AI models

Anthropic has been given a deadline of February 27 by Defense Secretary Pete Hegseth to grant the US military broad access to its AI models.

Hegseth warned that failure to comply could lead to Anthropic being labeled a “supply chain risk” or the Defense Production Act being invoked.

Negotiations have stalled over Anthropic’s concerns about restrictions on its models’ use, especially regarding autonomous weapons and mass surveillance.

The Department of Defense wants unrestricted access to all lawful uses, while Anthropic seeks assurances its models won’t be used for certain military applications.

A company spokesperson said discussions are ongoing to ensure responsible support for government security needs.

🔗 Source: CNBC

🧠 Food for thought

Implications, context, and why it matters.

The Pentagon’s ultimatum seeks wider rights to use Anthropic’s models, as dispute over ‘red lines’ is contested

The fight seems tied to Anthropic’s ethical rules for model use, including concerns about autonomous weapons and mass surveillance. A Pentagon official says the current clash has nothing to do with the company’s stated “red lines” on those topics 1.
The Department of Defense wants permission to use Anthropic’s models for “all lawful use cases.” It argues that buyers should control how purchased technology gets used 1.
Defense officials describe the tools as necessary, which drives a hard stance in talks. One official said, “The only reason we’re still talking to these people is we need them and we need them now. The problem for these guys is they are that good” 2.
Politics adds heat as some administration officials have criticized Anthropic in public. AI czar David Sacks, a White House adviser focused on AI policy, has attacked the company for supporting “woke AI” 3.

The ultimatum lays out government leverage over AI vendors, with ripple effects across the industry

The government has threatened to use the Defense Production Act (a U.S. law that can compel private companies to prioritize government needs). That approach treats a private AI company as a strategic asset that can be forced to comply on national security grounds 1.
The ultimatum raises the risk that federal demands override company limits on model use. Other Pentagon contractors, including Google, OpenAI and xAI, face the same DoD push for “all lawful use cases” 1.
The DoD has cleared Elon Musk’s xAI for classified networks, giving it a second approved provider alongside Anthropic. That change cuts Anthropic’s exclusivity and can tighten negotiations 3.
The pressure could lead AI firms to loosen ethical guardrails to win defense work. It may also speed up an AI arms race as rivals watch the U.S. expand military use of commercial AI.