AI generates covertly racist decisions about people based on their dialect - Nature
www.nature.com
Despite efforts to remove overt racial prejudice, language models using artificial intelligence still show covert racism against speakers of African American English that is triggered by features of the dialect.

Got the pointer to this from Allison Parrish who says it better than I could:

it’s a very compelling paper, with a super clever methodology, and (i’m paraphrasing/extrapolating) shows that “alignment” strategies like RLHF only work to ensure that it never seems like a white person is saying something overtly racist, rather than addressing the actual prejudice baked into the model.