Once again I just can't help but laugh at how my fellow engineers coddle AI.

"Hey, maybe social media shouldn't be made available to children. Let's add some age checks maybe?" "Reee noo that's invasion of privacy it's a slippery slope"

"Hey, my totally really, totally dangerous model, is too dangerous for adults. Let's add some checks maybe?" "Oh yes that sounds perfectly reasonable"

Who says that's perfectly reasonable? The first thing to ask any new LLM is how do you make cocaine, to figure out how nerfed it is.