It takes five minutes to “gaslight” a LLM into condoning genocide.

Also, there are several documented cases of AI chatbots involved, one way or another, in murders and suicides.

But more importantly, LLMs cannot have a moral compass. Saying otherwise is just wishful thinking.