AI’s biggest shift right now? Agents, security, and defense

AI just crossed a new line: it’s moving from chat to action, and the biggest players are signing up fast. Here are the must-know moves shaping the latest AI wave, from Pentagon deals to agent-native software and new security pressure.^[35][10]

Pentagon AI deal rush: the Defense Department says it struck agreements to put AI on classified networks at IL6 and IL7, with Oracle added to bring the total to eight. The goal is faster warfighting, intelligence, and enterprise ops.^[12]

Seven big tech firms, including OpenAI, Google, Microsoft, AWS, Nvidia, SpaceX, and Reflection, are now in the Pentagon mix. Anthropic is out after refusing the "lawful use" clause, turning AI governance into a real procurement fight.^[10]

AI is becoming agent-native. Fortune says Anthropic’s Mythos exposed serious governance risks, while the same week’s AI coverage points to agents that can plan, use tools, and execute multi-step work. The question is no longer "can it talk?" but "can it be trusted?"^[30][35]

Security is racing to catch up. SecurityWeek reports AI-powered exploitation is speeding up attacks, while Anthropic is pushing Claude Security and Cisco released an open-source tool for AI model provenance. The AI arms race is now defensive too.^[40]