Pentagon Sends Anthropic Final Offer for Military AI Use Amid Dispute

The Pentagon has sent Anthropic a best and final offer for military use of its Claude AI technology, with Defense Secretary Pete Hegseth setting a Friday evening deadline for the company to grant all lawful use or face losing its U.S. military business.

Key Details from the Negotiations

• Pentagon officials sent the offer Wednesday night, just ahead of a government-imposed deadline
• Defense Secretary Pete Hegseth demands a signed document granting full access to Anthropic's AI model by end of week
• Anthropic faces not just loss of business but being labeled a supply chain risk if it doesn't comply
• Pentagon is considering invoking the Defense Production Act to force compliance

Anthropic's Position and Concerns

• Anthropic has repeatedly asked for guardrails restricting Claude from mass surveillance of Americans
• CEO Dario Amodei wants to ensure Claude is not used for final targeting decisions without human involvement
• Company concerns include Claude's potential for hallucinations and unreliable behavior that could lead to lethal mistakes
• Anthropic was awarded a $200 million Pentagon contract in July to develop AI capabilities for national security

Military's Stance

• Pentagon officials say they're only asking for a license to use AI strictly for lawful activities
• Trump officials noted that mass surveillance is illegal and the Pentagon follows the law
• The military wants full control of AI technology for use in military operations

The outcome remains unclear as Anthropic spokespeople didn't immediately respond to requests for comment, and it's unknown whether the final offer substantially changed what the government has been seeking or whether the company has agreed.