The screening step where a red team's raw attack attempts are checked for whether they are genuine, reproducible failures worth reporting.
Continue to AI University →