Our Agreement with the Department of War

OpenAI just inked a deal with the U.S. military (the Pentagon, a.k.a. the Department of War) to run its AI in classified settings, promising big red lines: no mass domestic surveillance, no running killer robots, no high‑stakes decisions without a human. They say it’s cloud‑only, with OpenAI staff in the loop, and more safeguards than rivals. But the internet didn’t clap — it pounced. The line that set everyone off: “for all lawful purposes.” Critics read that as a legal escape hatch, not a moral stand. One commenter called it “not great” and “loose,” pointing out it doesn’t ban autonomous weapons — it just defers to whatever the rules say. Another asked the obvious: what if the rules change? OpenAI says the contract references today’s laws and policies, but skeptics weren’t buying it. The vibe in the threads: trust issues. People dragged OpenAI’s journey from nonprofit to for‑profit, saying every time a promise gets hard, it gets… revised. The meme machine went wild: “Terms of War,” “Skynet with a Terms of Service,” and “guardrails are vibes” were everywhere. A quieter camp argued this is the least‑bad option — better OpenAI with guardrails than someone else with none. But the loudest chorus accused the company of outsourcing morality to outdated laws, with “loopholes you could drive a tank through.” Read the post and bring popcorn.