大模型安全
- 大模型安全非常重要
- Fool the LLM by role play
- Jailbreak
- Jailbreak using base 64
- Jailbreak using Universal Transferable Suffix
- Jailbreak using adversarial examples
- Jailbreak using adversarial examples
- Prompt Injection via self-hosted webpages
- Prompt Injection via self-created docs
- Data Poisoning / Backdoor Attacks
- Security in LLMs
- Ending Pose