Court allows Pentagon to label Anthropic a national security risk; rogue OpenAI agents reportedly accessed US government sites · Digg