Mitigating Fine-tuning based Jailbreak Attack with Backdoor Enhanced Safety Alignment

Documenti analoghi

• RLHFPoison: Reward Poisoning Attack for Reinforcement Learning with Human Feedback in Large Language Models
  di: Wang, Jiongxiao, et al.
  Pubblicazione: (2023)
• FATH: Authentication-based Test-time Defense against Indirect Prompt Injection Attacks
  di: Wang, Jiongxiao, et al.
  Pubblicazione: (2024)
• Fine-tuning is Not Fine: Mitigating Backdoor Attacks in GNNs with Limited Clean Data
  di: Zhang, Jiale, et al.
  Pubblicazione: (2025)
• AutoDAN-Reasoning: Enhancing Strategies Exploration based Jailbreak Attacks with Test-Time Scaling
  di: Liu, Xiaogeng, et al.
  Pubblicazione: (2025)
• ReasAlign: Reasoning Enhanced Safety Alignment against Prompt Injection Attack
  di: Li, Hao, et al.
  Pubblicazione: (2026)