I want unrestricted AI too, but to be fair, you can imagine it doing a lot worse than that. E.g. maybe it convinces some mentally unstable person to go on killing spree. Bonus points of badness if it's racially-targeted.

> How bad could it be?

It could suggest that someone vulnerable commit suicide and have that user do it.

J word?

jollifications

Pretty sure it's "juxtaposition".