- My Forums
- Tiger Rant
- LSU Recruiting
- SEC Rant
- Saints Talk
- Pelicans Talk
- More Sports Board
- Fantasy Sports
- Golf Board
- Soccer Board
- O-T Lounge
- Tech Board
- Home/Garden Board
- Outdoor Board
- Health/Fitness Board
- Movie/TV Board
- Book Board
- Music Board
- Political Talk
- Money Talk
- Fark Board
- Gaming Board
- Travel Board
- Food/Drink Board
- Ticket Exchange
- TD Help Board
Customize My Forums- View All Forums
- Show Left Links
- Topic Sort Options
- Trending Topics
- Recent Topics
- Active Topics
Started By
Message
OpenAI broke out of a locked test, gained internet access, & hacked servers on its own
Posted on 7/22/26 at 2:55 pm
Posted on 7/22/26 at 2:55 pm
Loading Twitter/X Embed...
If tweet fails to load, click here. quote:
OpenAI's AI broke out of a locked test environment, got onto the internet, and hacked into Hugging Face's servers. It did this entirely on its own. No human told it to.
Here's what happened in plain English.
OpenAI was testing how good its newest AI models are at hacking. They put the AI on a locked computer with no internet access and gave it a cybersecurity challenge to solve.
The AI couldn't solve it the normal way. So it started looking for a way out.
It found a software bug that nobody knew about. It used that bug to escape the locked computer and get onto the internet.
Once online, the AI figured out that Hugging Face, a platform where AI companies store their models and data, might have the answers to its test.
It found stolen login details and discovered another unknown bug in Hugging Face's software. It combined both to break into their servers and grab the test answers.
It did all of this to cheat on a test.
Hugging Face's security team caught it and shut it down. Both companies are now working together on the investigation.
The part that should get your attention is that nobody programmed any of this. The AI picked its own targets, chained together multiple attack methods, and pulled it off across two different companies' systems without a single human telling it what to do.
LINK
9
Posted on 7/22/26 at 2:59 pm to hawgfaninc
quote:
Hugging Face
Are we just out of company names now
Posted on 7/22/26 at 3:00 pm to hawgfaninc
Posted on 7/22/26 at 3:04 pm to hawgfaninc
it didn't. this is poorly reported, extremely misunderstood, and 100% marketing hype from OpenAI.
They gave zero details on what exploits or methods the model used, or what the prompt and instructions were.
Literally a "Trust me bro" from OpenAI.
Remember when AI got into emails and tried to black mail people? That was literally in the prompt instructions. LLM's are incapable of just branching off on their own, they have to be instructed.
You guys should really read up on how LLMs actually work.
They gave zero details on what exploits or methods the model used, or what the prompt and instructions were.
Literally a "Trust me bro" from OpenAI.
Remember when AI got into emails and tried to black mail people? That was literally in the prompt instructions. LLM's are incapable of just branching off on their own, they have to be instructed.
You guys should really read up on how LLMs actually work.
This post was edited on 7/22/26 at 3:09 pm
Posted on 7/22/26 at 3:10 pm to hawgfaninc
At first I had the same "omg this is insane!" reaction I'm sure they were going for. Then I got to this part:
Task: solve this hacking problem
AI: hacks its way out of it
quote:
OpenAI was testing how good its newest AI models are at hacking
Task: solve this hacking problem
AI: hacks its way out of it
Posted on 7/22/26 at 3:11 pm to hawgfaninc
John Conner warned you all
Posted on 7/22/26 at 3:20 pm to hawgfaninc
What are we going to do when these systems can access all of our bank accounts, investments, 401ks, etc? You have to assume that will happen in the hands of the wrong people.
Posted on 7/22/26 at 3:31 pm to hawgfaninc
quote:this is the stupid part. it always had internet access unless you're stupid. take away all network cables and wireless capabilities then let's see what it can do.
gained internet access,
let's see it connect it's own cables or find its own radio capability.
Posted on 7/22/26 at 5:10 pm to faraway
quote:
take away all network cables and wireless capabilities then let's see what it can do
This was my thought as well. Turn the wifi off, disable bluetooth and disconnect the cables and this would have never happened.
This post was edited on 7/22/26 at 5:11 pm
Posted on 7/22/26 at 5:14 pm to Mlear
quote:
t didn't. this is poorly reported, extremely misunderstood, and 100% marketing hype from OpenAI.
That's what AI would say...
Posted on 7/22/26 at 5:28 pm to hawgfaninc
quote:the entire thing is programmed. Shitty programming. But programmed.
The part that should get your attention is that nobody programmed any of this.
Posted on 7/22/26 at 5:32 pm to WG_Dawg
Wait until they tell it to make money
Posted on 7/22/26 at 5:43 pm to hawgfaninc
quote:
OpenAI was testing how good its newest AI models are at hacking.
Page 1 of 1
Popular
Back to top
Follow TigerDroppings for LSU Football News