AI - it’s the future?

[jem]

Originally Posted by Carth

Don't you find it odd that a company at the forefront of AI development suddenly decides to 'release' a story about their AI going wrong during testing?

It's almost as though it's a ploy to get everyone talking about AI, and get it plastered all over the media.

Of course, using phrases like "some of its most advanced AI models went rogue" is quite obviously just a throw away line, not intended at all to get the internet all frothy about it

Well yes of course it is, but hypothetically as OpenAi have publically admitted that their system launched an attack against ‘Hugging Face’*, how about the latter sues the former for say $100 billion, or however much it takes to put them out of business.

Just a thought.

* wonderful name, wonder where it came from - cough Alien, cough!

[Carth]

** wonderful name, wonder where it came from **

swap the initial letters of Hugging & Face around, then say it fast

that's a *cough cough*

[OLD BOY]

Originally Posted by Carth

Don't you find it odd that a company at the forefront of AI development suddenly decides to 'release' a story about their AI going wrong during testing?

It's almost as though it's a ploy to get everyone talking about AI, and get it plastered all over the media.

Of course, using phrases like "some of its most advanced AI models went rogue" is quite obviously just a throw away line, not intended at all to get the internet all frothy about it

It doesn’t exactly show them in a good light, though, does it? I would have thought they’d want to keep that under wraps.

[damien c]

Originally Posted by OLD BOY

It doesn’t exactly show them in a good light, though, does it? I would have thought they’d want to keep that under wraps.

Problem is legally I think they have to declare it as it's technically a "hack/breach" even if it's by their own system.

What I don't understand is, if the system wasn't supposed to be able to have internet access, why the F did they have it connected to a network??

OpenAI had placed them inside what it called a “highly isolated environment,” with only limited access to an internal service used to download approved software. They found a previously unknown flaw in that service, used it to break into other OpenAI systems and eventually reached the open internet. From there, they inferred that Hugging Face might hold material related to the test, broke into its systems and obtained information that helped them score higher.

https://time.com/article/2026/07/24/...g-face-attack/