Generative AI at Work / Unit 7 GAI 152 short lesson, about 20 seconds
Memorization
Captions available
The idea
Models can memorize parts of their training data, and NIST notes that under adversarial attacks they have revealed sensitive information that was included in it.
In practice
A model prompted in a particular way reproducing a real person's contact details from a page it was trained on.
Where people slip
NIST adds that even data appearing only a few times in training can be at risk.
Your place on the trail
Every lesson rests on earlier ones and opens later ones. Tap any to jump.
Builds on
Unlocks
- Nothing yet depends on this.
Something wrong in this lesson? Flag it · Sources for Generative AI at Work