Emil Michael, the Pentagon's AI portfolio lead, designated Anthropic a supply chain risk after failed negotiations over military AI safety guardrails. The move has reshaped AI contracts with other major players.
TL;DR
- Emil Michael, former Uber executive, led Pentagon negotiations with Anthropic over military AI safety.
- Anthropic refused terms, leading to its designation as a supply chain risk—the first for an American company.
- OpenAI, xAI, and Google later agreed to Pentagon terms closer to Michael's original proposal.
What happened
Emil Michael, who joined the Pentagon's AI portfolio last August, attempted to negotiate new terms with Anthropic for military AI use. He sought to remove what he described as burdensome Biden-era restrictions that hindered basic military activities. According to Michael, he aimed to allow 'all lawful uses' of AI models.
Anthropic, however, refused the terms, citing two key exceptions: preventing Claude from conducting mass surveillance of Americans and from independently controlling lethal autonomous weapons. The company stated that the dispute was about these exceptions, not about inserting itself into the military chain of command, as Michael claimed.
After negotiations broke down, the Pentagon designated Anthropic a supply chain risk in February, the first time such a label has been applied to an American company. This designation came after Anthropic reportedly turned the negotiations into a public relations issue, according to Michael.
Why it matters
Michael's actions have significant implications for AI startups and investors. The designation of Anthropic as a supply chain risk sets a precedent for future negotiations between the military and AI companies. It highlights the tension between national security needs and corporate governance.
For developers and startups, the outcome underscores the importance of clear and acceptable use policies in AI contracts. The competitive landscape has shifted, with OpenAI, xAI, and Google agreeing to terms closer to Michael's original proposal. This could influence how other AI companies approach similar negotiations.
The dispute also raises questions about the role of AI companies in military operations. The debate over who controls the use of AI models in national security contexts is far from settled, and this incident may prompt further discussions and regulations.
Key facts
- Emil Michael led Pentagon negotiations with Anthropic over military AI safety guardrails.
- Anthropic refused terms, leading to its designation as a supply chain risk in February 2026.
- This is the first time an American company has been labeled a supply chain risk.
- Michael sought to remove 25 pages of restrictions from Biden-era contracts.
- Anthropic cited two key exceptions: preventing mass surveillance and autonomous weapons control.
- OpenAI, xAI, and Google later agreed to Pentagon terms closer to Michael's proposal.
- Michael previously led negotiations for Uber's $3.5 billion Saudi investment and a deal with Didi in China.
- Anthropic investors include Salesforce, where TIME owner Marc Benioff is CEO.
Context
This incident occurs amid growing concerns about the ethical use of AI in military applications. The designation of Anthropic as a supply chain risk marks a significant escalation in the tension between national security interests and corporate governance.
The broader AI industry is watching closely, as the outcome could set precedents for future negotiations and regulations. The competitive dynamics between major AI players are also at play, with OpenAI, xAI, and Google aligning with the Pentagon's revised terms.
The role of AI in national security is a complex and evolving issue. This dispute highlights the need for clear guidelines and policies that balance the needs of the military with the ethical considerations of AI development and deployment.
