Comment by hashmap

kind of boggles me that the big ai labs arent running zero-shot classification on all their outputs with "does this look like any of these bad things we know people get up to"