Anthropic Rejects Latest Pentagon Offer, Escalating AI Feud

Anthropic, the AI safety-focused company behind the Claude models, has publicly declined a major contract offer from the U.S. Department of Defense, marking a significant escalation in the ongoing debate over ethical AI development for military applications. The company stated the decision was based on its 'Long-Term Benefit Trust' structure and constitutional commitment to avoid enabling 'catastrophic risks,' even as the Pentagon seeks to integrate advanced AI for intelligence analysis, logistics, and cyber operations. This refusal places Anthropic alongside a small but vocal group of AI labs, including some OpenAI alumni, who are drawing hard ethical lines against weaponized or high-stakes government AI systems.

This rejection occurs against a backdrop of rapidly increasing defense AI investment, with the Pentagon's Joint Artificial Intelligence Center overseeing a budget that has grown to over $1.8 billion. The move intensifies a strategic and philosophical feud within the tech sector, pitting companies with strong safety mandates against both government agencies and rival AI firms willing to engage in defense work. For the Pentagon, the refusal complicates efforts to modernize with cutting-edge large language models (LLMs) and autonomous systems, potentially ceding technological ground to adversaries. The standoff signals a new phase where corporate AI ethics policies are directly impacting national security procurement, forcing a reevaluation of how advanced AI is sourced and governed for public sector use.