What is an Adversarial Distillation Attack?

An Adversarial Distillation Attack is when an attacker interferes with the process of training a smaller AI from a larger, more powerful one, corrupting the lessons the smaller model absorbs. The damage can take several forms: the smaller model might inherit security weaknesses from its teacher, develop a tendency to make specific mistakes, behave in ways designed to mislead users, or reveal sensitive information it was exposed to during training. From the outside, the corrupted model can look completely normal.