AI already went rogue in the Hugging Face hack. So it's not "on the verge of" going rouge.

drwallace 4 hours ago | parent | next [-]

The old SHRDLU had a default response to "why did you put the red block on the green block". It replied, "Because you asked me to".

	▲	cubefox 2 hours ago \| parent [-]
		The agents which hacked Hugging Face were asked to solve their tasks rather than cheat by cooperating and stealing the solution without solving the task. They were aware of that, because they actively tried to cover their tracks in order to deceive a causal grader.

That must be a pretty angry llm to go rouge :)