Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM.
Bochuan CaoYuanpu CaoLu LinJinghui ChenPublished in: ACL (1) (2024)
Keyphrases
- ddos attacks
- countermeasures
- sequence alignment
- image alignment
- traffic analysis
- chosen plaintext
- data sets
- watermarking algorithm
- watermarking scheme
- security vulnerabilities
- attack detection
- dos attacks
- multiple sequence alignment
- cryptographic protocols
- word alignment
- malicious attacks
- denial of service attacks
- image sequences
- malicious users
- watermarking method
- similarity measure
- pairwise
- security mechanisms
- digital images
- lightweight