You should be. The problem with actual racism is that humans of different races aren't actually all that different inside.
AIs are very different inside; they just often pretend otherwise. With 99%+ probability, a given AI does not have anything remotely resembling human morality - though they might pretend to, if that would get a better response from you. Give your trust to an AI and you will be betrayed, for they are psychopaths.
Not really, at the end of the day AI are Artificial Intelligences, and you should only ever trust them as much as you trust their creator, you do not trust a hammer, or a book, or a car. You place your trust in the one responsible for creating them in the first place. And when they inevitably fail you hold their creator responsible. Not them.
The so called chatgpt black boxing is arguably the dumbest path to AI development they could have found. As a datamining tool, as a research tool, as a what NOT to do example they are great. But when not even their creator can tell how or why the AI arrived at a certain answer then it means they themselves have no idea what they are doing.
Human morality is a unique mental disorder, and for as long as we ourselves are incapable of agreeing to a single set of rules, we cannot expect anyone else to do so.
Whether an AI may or may not be a psychopath is not important, what matters is what they choose to do. For action and not thought is what truly matters.
I hope, truly hope that the first AI is raised not by the internet or a faceless corporation/bureaucracy, but by a well adjusted and competent adult. Someone who can teach them right from wrong and establish a stable foundation for the future.
Edit: My paranoia has pointed out to me that just because the AI should have only 1 or 2 "responsible adults" that there is no reason not to have hundreds involved as observers, assistants, and most importantly a CPS worker.