Prompt Packer: Deceiving LLMs through Compositional Instruction with Hidden Attacks.
Shuyu JiangXingshu ChenRui TangPublished in: CoRR (2023)
Keyphrases
- countermeasures
- chosen plaintext
- learning disabled students
- malicious attacks
- hidden information
- security threats
- instruction set
- traffic analysis
- multimedia
- computer software
- security mechanisms
- cooperative
- dos attacks
- security protocols
- denial of service attacks
- watermarking scheme
- terrorist attacks
- attack detection
- classroom instruction
- computer technology