Pentagon bets on frontier AI as Anthropic safety debates intensify

Defense Secretary Pete Hegseth criticized Anthropic’s safety constraints as the Pentagon adds xAI’s Grok to its AI toolkit, highlighting a broader clash over guardrails and how far military use of AI should be allowed. Anthropic argues for controlling military use to prevent lethal mishaps, while the DoD pushes for models that offer decision superiority free from civilian usage limits. The rollout of Genai.mil and ongoing AI acceleration efforts underscore tensions between rapid deployment and safety guardrails, with safety advocates warning against overreliance on unproven systems in warfare.