LLMs tend to memorize information including PI information leading to privacy issues. Machine Unlearning is a new setup wherein a pretrained model is made to “forget” certain information regarding some users who request their data to be removed.
Abhinav Joshi, Shaswati Saha, Divyaksh Shukla, Sriram Vema, Harsh Jhamtani, Manas Gaur, Ashutosh ModiFindings of the Association for Computational Linguistics: EMNLP 2024
Author 1, Author 2