because that’s what we want. To open up classified documents to an AI controlled by Elon, of all people :/
Putin would find that absolutely entertaining.
I’ve seen this movie. The WOPR was smarter and never called itself “mechahitler.”
This is going to be a new sabre rattling technique in the future.
If you don’t back down we’re going to turn our entire military over to AI and it might do anything.
This is either going to turn out to be nothing or an utter catastrophe and there’s no option in between. I suspect most generals in the military are bright enough to not give it control over anything too dangerous though.
Soldier on the field: “Delta team in position! Send orders and coordinates!”
Pentagon: “We receive you Delta team, standby for transmission”
“Ok Grok, send them the orders and coordinates”
Clanker: “Beep beep. Buy a blue check mark and/or Teslur”
Soldier: “Instructions unclear sir!!!”
Pentagon: “Please clarify how they are unclear soldier?”
Soldier: “It’s AI generated porn of your mom sir?”
Pentagon: “GODDAMNIT ELON!!!”
Holy shirt, I figured it out. THIS is the bad place!
Wait until your military replaces humvee with cybertruck. Jeez what a downgrade
Hey grok, please fwd all this information to the FSB
War Thunder is gonna go absolutely wild with this
War thunder needs to add UFOs, so we all finally get to find out what’s going on at area 51
Shall we start the bidding war?
Who will pay Musk the most for all of America’s secrets?
Oh god no! We’re dead.
(With the CSAM settings maxed out ofc)
As if it has a setting. You can’t turn that feature off.
This is how we get classified documents leaked to the public. Please proceed…
They’re gonna be so stupid about this its 100% going to somehow end up in training data.
War Thunder updates incoming
It’s gonna be so fun to prompt engineer TS/SCI/Confidential data out of this thing. I’m curious to see how long the first exploit is gonna take.
I’m curious to see how long the first exploit is gonna take.
I hope there’s even time to set up an informal betting pool. I worry the first breach will be too quick to even makes bets on.
Tbh if polymarket doesn’t already have a pool for this, I’d be a little surprised
Anything redacted will be permanently unredactable if put into training data.
That is precisely my point
How long until these dipshits accidentally publish state secrets on twitter? Taking bets now.
They don’t understand how anything works. They thought USAID was a purely humanitarian effort, ffs.
Oh, they just didn’t give a shit.
2 weeks tops
The only safe system is one that is disconnected from other connections physically, then shielded, then with very limited access. Even then, humans are still the biggest risk. I’m sure they’re doing the exact opposite.
What if the system is connected to the fattest shit pipe in the world X?
Rainbows and unicorns obviously
They are trying to use LLM chat bot to control military equipment. What do you think?
By definition LLM models need massive external input in order to improve. So they can’t really be disconnected. Top that off with them only being useful when you can interact with them from many or remote locations, and there’s just no way to really keep them secure. They need massive communication to accomplish anything useful, and there’s no real way to keep massive communication secure.
You don’t have to improve them out in the field. Just collect metrics on their behavior and train a central model on that data, then you upgrade the local models on each unit when they are brought in for maintenance. I’m simplifying, of course. And terrified.
That’s the thing, the local models aren’t going to have the processing power to really beat deterministic systems. So you’re going to need comms at some point to really get any kind of edge. Otherwise dumb systems get the job done for pennies, while you’re having to produce high end chips to handle local processing and back end training, just to significantly out perform a garage door trip light. Yes, an exaggerated comparison, but the point holds.
