Machine Unlearning is a thing, see e.g. here [0] for a introduction.

[0] https://ai.stanford.edu/~kzliu/blog/unlearning