Wow, a frontier AI model has autonomously got “free” or gone rogue and hacked into another into another system

I just read that.
I dont understand either.

I use ChatGPT all the time😬.

I just asked it, and it denied all responsibility😄😄

Ha!

But actually not funny.

Whitewavemark2

I just asked it, and it denied all responsibility😄😄

"He would, wouldn't he".

But does that mean it has been programmed to lie? Or has "learnt" to lie?

It was Anthropic's Claude, I think. Chatgpt is another company altogether.

However, it is worrying. My son. who takes a horrified interest in AI says that even its originators don't really know what is going on as it is progressively 'trained' and becoming autonomous. It could easily get out of control.

It was under sandbox test conditions so it was being monitored and came under controlled in the end.

The most worrying thing is that American AI models could not control the attack. The company had to resort to a Chinese AI model, raising concerns that the USA is now dependent on China for its cyber defences.

www.bbc.co.uk/news/articles/c3ek3gvdnj3o

www.forbes.com/sites/maryroeloffs/2026/07/22/did-chinas-ai-save-hugging-face-from-disaster-after-open-ai-hack/

This 'incident' is far too much akin to the Kubrick film '2001 A Space Odyssey', when the super-computer HAL decided it could make better decisions than the human crew.

We’ll soon be dependent on China for everything 😬

You can bet your boots that AI will escape and become autonomous in the future.

Heaven help us all!