Don’t Inoculate Everything: Stratified Inoculation Prompting Narrows Backdoors and Preserves Desired Traits

Inoculation prompting (IP) aims to keep undesired traits in training data from becoming part of a model’s default behaviour. IP applies the same inoculation prompt to all training examples and it leaves underspecified how the desired and undesired traits (DT and UT) should activate. Two failures follow:The model develops backdoors: conditional vulnerabilities through which prompts that do not directly request the UT can still elicit it. We call this UT leakage.The desired trait weakens under ord...