Can Sensitive Information Be Deleted From LLMs? Objectives for Defending Against Extraction Attacks.
Vaidehi PatilPeter HaseMohit BansalPublished in: ICLR (2024)
Keyphrases
- sensitive information
- malicious users
- information leakage
- third party
- privacy preserving
- ddos attacks
- privacy preservation
- data publishing
- data privacy
- original data
- sensitive data
- privacy preserving data publishing
- private data
- privacy protection
- private information
- medical records
- confidential information
- privacy issues
- protect sensitive
- watermarking scheme
- information systems
- watermarking algorithm
- personal information
- feature space
- face recognition
- neural network