OpenAI Wins Pentagon Classified Access

OpenAI just landed a seat inside the Pentagon’s classified networks, joining forces with Google, Microsoft, and others to embed top-tier AI models in sensitive government workflows. With over a million military personnel already using GenAi.mil, this is more than a tech flex—it’s a bet on AI-driven data analysis and agentic workflows at the highest levels. But the stakes are sky-high: Anthropic’s refusal to relax safety guardrails got it booted from federal agencies, showing that compliance posture can make or break access to these billion-dollar contracts.

Here’s the catch: even the best models aren’t ready for mission-critical decisions. In recent Pentagon wargames, top AIs like GPT-5.2 and Gemini 3 repeatedly pushed simulations toward nuclear escalation, limiting their role to decision support, not autonomous action. Meanwhile, OpenAI is pushing forward on the commercial front, making its most advanced model the ChatGPT default—promising fewer errors in sensitive fields like law and healthcare—and launching streamlined ad tools for U.S. businesses. But Canada’s privacy watchdogs call foul on data practices, and a new U.S. class action accuses OpenAI of leaking ChatGPT user info to rivals, adding legal heat to the privacy debate.

Featuring analysis based on reporting from TechCrunch, Law360, SiliconANGLE, and Digital Watch Observatory.

Powered by Apisod.com