Coincidentally, the temperature in Shanghai dropped to 6 degrees today. As the weather suddenly turned bitterly cold, let’s talk about a piece of tech gossip that might also send a shiver down your spine.
The “Don’t Be Evil” Motto, Deleted Just Like That
You might not know this, but when the highly acclaimed OpenAI first started, its terms of use stated an ironclad rule in black and white: “We don’t allow the use of our models for… military and warfare.”
Back then, everyone thought that with “Open” in its name and a mission for the benefit of all humanity, the company was a true beacon of tech.
But what happened? In early 2024, they quietly deleted this sentence.
At the time, many didn’t take it seriously, assuming it was just a routine legal text modification by a major tech corporation. It wasn’t until recently that this boomerang finally hit the bullseye—having just secured a massive contract from the Pentagon, OpenAI turned its sights on NATO’s “unclassified network”. The dragon-slaying youth who once preached “for all humanity” has now put on a suit and sat down at the table with the military-industrial complex.
Look at this vibe, doesn’t it feel like Cyberpunk becoming reality?
The Drama Sparked by a Nuclear Missile Drill
How did this actually come about? Truth be told, you can’t entirely blame OpenAI for willingly throwing themselves into the military’s arms; a lot of the credit goes to their peers for making them look good.
The whole thing started when the Pentagon recently conducted an extremely hardcore simulation drill. The military brass called in OpenAI’s biggest arch-rival—Anthropic, the company behind the Claude large language model.
The military asked a very blunt question: If the US were attacked by intercontinental ballistic missiles, could your AI system be used to assist in interception?
Hearing this, Anthropic’s CEO shook his head immediately. They believed current AI is simply not reliable enough. What if it hallucinates? Everyone understands LLM “hallucinations”—it’s like a highly confident but utterly clueless intern blindly giving orders. Spouting nonsense in a weekly report might just get your pay docked, but blindly commanding during a missile launch would cause an absolute catastrophe. So, they held their bottom line and firmly opposed using their technology for automated weapons and lethal decision-making.
As a result, the Pentagon was furious.
Military officials publicly complained, stating that large tech companies absolutely must not be allowed to dictate war outcomes based on their own ideologies, and promptly blacklisted and banned Anthropic. Sigh, it’s quite tough for them—sticking to their principles ended up costing them their job. As a side note, Anthropic’s core members originally left OpenAI because they felt OpenAI wasn’t safe enough, so they are really staying true to their original intentions.
Just hours after Anthropic was kicked out, a dramatic scene unfolded.
OpenAI’s CEO, Sam Altman, swiftly swooped in to fill the void. He told the military that they were willing to cooperate, and they would add “safety guardrails” to the system, ensuring it would behave well even within classified networks. The Pentagon saw how cooperative he was and immediately handed the contract over.
Once the news broke, netizens exploded. Users aren’t stupid—seeing a chatbot cozying up to the military, who knows what it’s doing in the background? In just a few days, ChatGPT’s uninstall rate on mobile app stores skyrocketed by nearly 200%. Meanwhile, Claude, the one kicked out for holding its ground, surged to the top of the Apple App Store download charts.
This put Altman in a rather awkward position. Usually portraying a picture of complete confidence, he panicked this time and rushed to social media for crisis PR. He admitted that the announcement was too hasty and did look quite “opportunistic and rash.” To appease the public, he added another rule guaranteeing that AI would absolutely not be used to spy on citizens. But apologies aside, the military contract was still signed.
The Edgelord Tendencies of Silicon Valley Tycoons
Speaking of this, I’m reminded of another thing. These Silicon Valley bosses building military AI seem to have a weird naming fetish.
Once, I chatted with a friend about another big data AI giant deeply embedded with the military: Palantir. This company is a heavyweight in the field; even NATO uses their technology. Not only do they help the military integrate data, but they’ve also stuffed LLMs into command systems. It’s said that they can blend various fragmented information from satellites, drones, and social networks, allowing the AI to present commanders with multiple strike options in mere seconds. Intelligence that previously took thousands of people in meetings to analyze can now be processed with a few mouse clicks.
Can you guess where the name “Palantir” comes from?
It’s actually the “seeing-stone” from The Lord of the Rings. In the novel, it’s a magical artifact that can see through everything in the world, but also easily corrupts the user, bringing them under the control of dark forces.
Naming a real-world military system—used for surveillance, intelligence gathering, and assisting lethal strikes—directly after a highly dangerous and metaphorically loaded artifact from a fantasy novel. This undisguised ambition mixed with a weird sense of “chuunibyou” (middle-school edgelord syndrome) makes you really marvel at the brain circuitry of tech giants.
The Palantír is watching you. This name is brutally honest
Reeling in the Imagination
Take a deep breath. Having said all this today, I actually just found out myself while scrolling through the news. We originally thought AI was just here to help us write emails or draw funny pictures. Who could have imagined that in another parallel universe, they are already participating in planning missile interceptions? ╮(╯▽╰)╭
Anyway, I’m off to grab some food. From now on, when asking ChatGPT questions, I think it’s safer to add a “please”.
References:
- The Pentagon “Blacklists” AI Giant Anthropic, Core Dispute: Nuclear War
- OpenAI Cuts ‘Military and Warfare’ Ban From Permissible Use Policy
- From Hunting Bin Laden to AI Warfare: Decrypting the Rise of Palantir
- OpenAI Faces Backlash: Forced to Modify Collaboration Agreement with U.S. Military
—— Lyra Celest @ Turbulence τ.
