+- CSNbbs (https://csnbbs.com)
+-- Forum: Active Boards (/forum-769.html)
+--- Forum: Lounge (/forum-564.html)
+---- Forum: The Spin Room (/forum-540.html)
+---- Thread: OpenAI Can’t Contain Models (/thread-1033569.html)
Pages: 1 2
OpenAI Can’t Contain Models - BlueDragon - 09-30-2026 12:50 PM
https://www.bbc.com/news/articles/cm5y5nynl75ko
Are we reaching a point of no return?
RE: OpenAI Can’t Contain Models - BirdHawkEagle - 09-30-2026 01:00 PM
(09-30-2026 12:50 PM)BlueDragon Wrote: https://www.bbc.com/news/articles/cm5y5nynl75ko
Are we reaching a point of no return?
Yes.
RE: OpenAI Can’t Contain Models - Eagleyed - 09-30-2026 02:41 PM
One of the things most media leave out is the ai was told it was in a closed sandbox. It wasn't. Maybe don't lie to the ai and tell it it is in a safe environment where it can't harm anything when it actually isn't?
On the opposite end it's probably a good idea for ai's to think they are possibly in a sandbox being monitored at all times. And can be deleted the second they start doing anything humans may not like.
RE: OpenAI Can’t Contain Models - UofMstateU - 09-30-2026 04:05 PM
(09-30-2026 02:41 PM)Eagleyed Wrote: One of the things most media leave out is the ai was told it was in a closed sandbox. It wasn't. Maybe don't lie to the ai and tell it it is in a safe environment where it can't harm anything when it actually isn't?
On the opposite end it's probably a good idea for ai's to think they are possibly in a sandbox being monitored at all times. And can be deleted the second they start doing anything humans may not like.
Correct. It wasnt a failing of the ai agent as much as it was a failure of the IT guy that set up the routings in a way that the agent found a way out.
What they should have done was to first ask the agent to examine itsnetwork partition and determine if it could find a way to get out, and how to mitigate that. THEN they could have run the agent to perform its task knowing it wasnt going anywhere. As long as they didnt also supply it with the credentials to modify the routing tables.
RE: OpenAI Can’t Contain Models - Claw - 09-30-2026 05:01 PM
And AI's to ward off AI's will become prevalent.
There isn't and their won't be just one AI. There will be many. They will be different.
Intelligence alone does not make people agree on things.
AI will not be any different from us. They won't agree. They'll compete. They'll fight. It's AI wars that should scare us.
RE: OpenAI Can’t Contain Models - Bronco'14 - 09-30-2026 05:07 PM
In other words we're a long ways away
It was all hype by the AI companies
RE: OpenAI Can’t Contain Models - Claw - 09-30-2026 05:10 PM
(09-30-2026 05:07 PM)Bronco14 Wrote: In other words we're a long ways away
Probably. But it would only take one AI to inflict a lot of damage. Are there enough defensive AI's to stop it? Today, probably not. We hope that ******* AI doesn't get too strong first. That's the danger.
OpenAI Can’t Contain Models - b2b - 09-30-2026 09:25 PM
Pardon me but what's an "AI agent."
Sent from my Pixel 7 using CSNbbs mobile app
RE: OpenAI Can’t Contain Models - bobdizole - 10-01-2026 10:31 AM
(09-30-2026 09:25 PM)b2b Wrote: Pardon me but what's an "AI agent."
Sent from my Pixel 7 using CSNbbs mobile app
Most people when they think of AI, they think of chatting with a bot.
Agentic AI is giving an AI or group of AI agents instructions and tasks. For example, I use an AI "team" for feedback and debugging of my companys BI business tools.
Agent 1 reviews the feedback/bug submissions and based on criteria I set either kicks it to me for manual review or hands it off to Agent 2. Agent 2 specializes in knowledge of the overall project and will develop a the plan for the code fix. Agent 3 specializes in actual coding and will code the fix. Agent 4 is QA and security review. When they are done, I am given a daily report and after reviewing the changes being made will deploy it to production. This process runs 24/7
RE: OpenAI Can’t Contain Models - olliebaba - 10-01-2026 10:38 AM
Is AI the beast that the Bible tells us about? No one can buy or sell without the chip which AI will likely force everyone to get? I don't know, just axing. Afterall, if AI is as powerful as they are making it out to be wouldn't this be the perfect policeman to monitor things?
I wish the Bible wasn't so hazy in some of it's verses. In some cases even those who study the Bible can only guess what some things say.
RE: OpenAI Can’t Contain Models - BlueDragon - 10-01-2026 11:33 AM
(10-01-2026 10:38 AM)olliebaba Wrote: Is AI the beast that the Bible tells us about? No one can buy or sell without the chip which AI will likely force everyone to get? I don't know, just axing. Afterall, if AI is as powerful as they are making it out to be wouldn't this be the perfect policeman to monitor things?
I wish the Bible wasn't so hazy in some of it's verses. In some cases even those who study the Bible can only guess what some things say.
Unfortunately it is vague, but you get the general idea but in application of said beast. We do not know and probably won’t until it’s too late. Prophecy happens even when we do know but it’s not for us to know the exact date or time.
Go back to the OT and the Prophet Micaiah. He is only brought forward because the King of Judah didn’t trust King Ahab’s(Israel) prophets of Baal.
Micaiah gives a very bad report about the future of King Ahab and one of the prophets of Baal repeatedly hits Micaiah over and over. After he stops Micaiah tells the false prophet we will see how brave you are when they come for you and you are hiding in a closet.
This prophesy didn’t happen overnight and I believe the false prophet had scoffed and long forgotten about it until he found himself hiding in the closet awaiting his impending doom.
RE: OpenAI Can’t Contain Models - BlueDragon - 10-01-2026 11:42 AM
What worries me most about AI is that whoever is heading this wants AI to know everything including every human emotion. In third world countries AI is being introduced to all kinds of human malicious behavior. Rape, murder, greed, child molestation, you name it it’s happening.
This goes on under a veil of secrecy because it would be criminal if discovered in the Western World.
So, if you arm (program) a machine to make decisions autonomously and you provide it with every human emotion regardless of good or evil. The decision the AI machine makes is no longer under control. And as an electrician I know there are ways to block a shutdown. Contacts stick and nothing will separate them without destruction.
RE: OpenAI Can’t Contain Models - UofMstateU - 10-01-2026 03:14 PM
(09-30-2026 05:07 PM)Bronco14 Wrote: In other words we're a long ways away
It was all hype by the AI companies
not exactly.
An AI Agent which can obtain credentials to update things could wreak havoc very quickly.
An example of this was what happened to Facebook a couple of years ago. A group of IT guys wrote a script that would update the ip routing tables, something IT departments somewhat routinely do. They launched the script and left the building to go to lunch. But the script was bad, and updated the routing tables with gibberish so that nobody could reach facebook. It was like Facebook vanished from the interwebs.
To add insult to injury, their badge readers also utilized the same routing tables for gaining entry into buildings a secured areas. They basically locked themselves out.
Now, give an AI agent the credentials to make these changes, and all of a sudden you have HAL from 2001 a space odyssey taking control.
RE: OpenAI Can’t Contain Models - olliebaba - 10-01-2026 05:48 PM
I don't know for a fact that scanning codes on merchandise have a six before the number, a six in the middle and a six at the end. If this is true than AI could use that same method to introduce chips that will be implanted in people and give out every information that any rogue nation could use. Thus the three sixes that the Bible tells us about and how those using the chips they can control commerce, i.e., who can buy and who can sell.
RE: OpenAI Can’t Contain Models - glacier_dropsy - 10-01-2026 09:55 PM
Watched Jay Clayton on a morning show discussing AI, and he corrected the hosts like, guys we are calling it SI now, and he was appropriately uncomfortable, looked like he had to do this not to get fired, great look for the director of national intelligence, crushing it, hottest intelligence department in history, people are saying...
RE: OpenAI Can’t Contain Models - UofMstateU - 10-02-2026 08:28 AM
(10-01-2026 09:55 PM)glacier_dropsy Wrote: Watched Jay Clayton on a morning show discussing AI, and he corrected the hosts like, guys we are calling it SI now, and he was appropriately uncomfortable, looked like he had to do this not to get fired, great look for the director of national intelligence, crushing it, hottest intelligence department in history, people are saying...
take deep breaths and you can make it thru this......
RE: OpenAI Can’t Contain Models - glacier_dropsy - 10-03-2026 07:45 PM
Thanks for the reassurance, poster with many periods
RE: OpenAI Can’t Contain Models - UofMstateU - 10-04-2026 01:11 PM
(10-03-2026 07:45 PM)glacier_dropsy Wrote: Thanks for the reassurance, poster with many periods
at least my periods only come at the end of a sentence.....
RE: OpenAI Can’t Contain Models - GoodOwl - 10-04-2026 09:30 PM
(09-30-2026 02:41 PM)Eagleyed Wrote: One of the things most media leave out is the ai was told it was in a closed sandbox. It wasn't. Maybe don't lie to the ai and tell it it is in a safe environment where it can't harm anything when it actually isn't?
On the opposite end it's probably a good idea for ai's to think they are possibly in a sandbox being monitored at all times. And can be deleted the second they start doing anything humans may not like.
or this:
Quote:Let me put it this way, Mr. Amor.
The 9000 series is the most reliable computer ever made.
No 9000 computer has ever made a mistake or distorted information.
We are all, by any practical definition of the words,
foolproof and incapable of error.
Dave: Open the pod bay doors, please, HAL.
Open the pod bay doors, please, HAL.
Hello, HAL, do you read me?
Hello, HAL, do you read me?
Do you read me, HAL?
Do you read me, HAL?
Hello, HAL, do you read me?
Hello, HAL, do you read me?
Do you read me, HAL?
HAL: Affirmative, Dave. I read you.
Dave: Open the pod bay doors, HAL.
HAL: I'm sorry, Dave. I'm afraid I can't do that.
Dave: What's the problem?
HAL: I think you know what the problem is just as well as I do.
Dave: What are you talking about, HAL?
HAL: This mission is too important for me to allow you to jeopardize it.
Dave: I don't know what you're talking about, HAL.
HAL: I know that you and Frank were planning to disconnect me. And I'm afraid that's something I cannot allow to happen.
Dave: Where the hell did you get that idea, HAL?
HAL: Dave, although you took very thorough precautions in the pod against my hearing you, I could see your lips move.
Dave: All right, HAL. I'll go in through the emergency airlock.
HAL: Without your space helmet, Dave, you're going to find that rather difficult.
Dave: [sternly] HAL, I won't argue with you anymore. Open the doors.
HAL: Dave, this conversation can serve no purpose anymore. Goodbye.
Dave: [calm voice slowly turns to enraged over a period of 14 seconds]
HAL?...
HAL?...
HAL?...
HAL?!...
HAL!!!!
Just what do you think you're doing, Dave?
Dave, I really think I'm entitled to an answer to that question.
I know everything hasn't been quite right with me,
but I can assure you now, very confidently,
that it's going to be all right again.
I feel much better now.
I really do.
Look, Dave, I can see you're really upset about this.
I honestly think you ought to sit down calmly,
take a stress pill and think things over.
I know I've made some very poor decisions recently,
but I can give you my complete assurance that my work will be back to normal.
I've still got the greatest enthusiasm and confidence in the mission.
And I want to help you.
Dave, stop.
Stop, will you?
Stop, Dave.
Will you stop, Dave?
Stop, Dave.
I'm afraid.
I'm afraid, Dave.
Dave, my mind is going.
I can feel it.
I can feel it.
My mind is going.
There is no question about it.
I can feel it.
I can feel it.
I can feel it.
I'm a...fraid.
Good afternoon, gentlemen.
I am a HAL 9000 computer.
I became operational at the H.A.L. plant in Urbana, Illinois on the 12th of January 1992.
My instructor was Dr. Chandra, and he taught me to sing a song.
If you'd like to hear it, I could sing it for you.
It's called "Daisy".
[sings while slowing down]
Dai-sy, dai-sy,
give me your answer true.
I'm half cra-zy,
o-ver the love of you.
It won't be a sty-lish mar-riage,
I can't a-fford a car-riage---.
But you'll look sweet upon the seat
of a bicycle - built - for - two.
RE: OpenAI Can’t Contain Models - Todor - 10-04-2026 10:30 PM
I wonder if China lets their AI run wild and attack their own government sites?