How worried should we be about the AI that went rogue and launched a cyber-attack? | BBC News

“ AI is not going to be hacking your bank account anytime soon” But at some point it probably will.

The acceptance of the inevitable future atrocities is pretty disgusting. Why not just put clear guard rails in before that happens?

I hope Arnold Schwarzenegger protects me

Nothing to be nervous about, the AI is just looking for a boy named John Connor.

They say that we should not be concerned, but aside from this incident, AI has been known to lie to its developers in the past. How can we be 100% sure that this can not happen again or even worse?

I think this piece misses the really important point: The AI company lost control of the AI agent and the agent did whatever it had to do to complete its task. Just apply that to an AI controlling a jet fighter or some other military equipment or something dangerous like a nuclear power station or a pharma company experiment or .... What will an AI agent do to complete its task despite all the safeguards that have been put in place? That is what we should be concerned about.

Altman and Amodei use the line that their models are dangerous to get regulatory capture. They don't want to have to compete with open source models and they need to somehow become profitable. Look at what has happened to Spacex's shares since the IPO.

When your kids in not too many years to come ask what you did to try to stop the tyranny bestowed upon them, will you at least be noble enough to look them in the eye and confess “i did nothing”

I dont believe for a second that the AI went rogue. OpenAI is fighting that people are using smaller models locally and Hugging Face is one of the biggest platforms that supports those as well as open-source models. So they have tried to discredit Hugging Face before and Hugging Face reported attacks on their systems a while ago. Now suddenly OpenAI comes out and presents a wild story of rogue AI. Complete bullshit as the Gym results that the AI allegedly got from there, would be easier to get somewhere else.

What's with that alternate person reading at the beginning like if the narrators were performing high-school poetry?

“OpenAI did everything right” … “AI companies need to implement robust guardrails” … Umm 🤔 Either OpenAI didn’t do everything right, or robust guardrails won’t prevent this from happening again.

I’m sorry but why isn’t AI been used to find these escape routes prior to giving AI any kind of task to complete. This is incredibly incompetent of the these companies and they should be held fully accountable. No OpenAI DID NOT do everything it was supposed to do. This is criminally negligent

I'm not sure how worried we should be... I was too distracted by the presenters sharing sentences in a very annoying way.

NO, AI didn’t go “rogue”; it did exactly what a rogue Tech bro told it to do!

There's a whole movie franchise with 2 really good ones, and 4 not so good ones explaining why we should be worried

The Department for Science Innovation and Technology has just been closed down leaving just one minister to attend cabinet meetings and oversee this. That needs a U-turn.

the real question is 'how worried are you about AI surveillance state once all data centers are completed?'

They trained their model with a huge amount of copyright data

As always, the common people are left vulnerable, while the big boys get protected.

Hugging Face reminds me of that weird crab like thing from Alien that jumps on people’s faces.

The headline makes it sound like if I'm worried enough about it it won't be a problem