Policy & Ethicsllmdefenseanthropicdata poisoning
Pentagon Integrates Commercial LLMs Into Defense
7.5
Relevance Score
The Pentagon accelerated integration of commercial AI models after awarding contracts last summer to OpenAI, Anthropic, Google, and xAI, with xAI’s Grok going live on classified systems in late January 2026. Reports link Anthropic’s Claude to the operation that captured Venezuelan leader Nicolás Maduro, spurring a government dispute and potential supply-chain penalties. Experts warn these consumer-oriented LLMs pose ethical, security, and data-poisoning risks in defense contexts.


