Jailbreak Antidote: Runtime Safety-Utility Balance via Sparse Representation Adjustment in Large Language Models

Documenti analoghi

• Harnessing Task Overload for Scalable Jailbreak Attacks on Large Language Models
  di: Dong, Yiting, et al.
  Pubblicazione: (2024)
• PandaGuard: Systematic Evaluation of LLM Safety against Jailbreaking Attacks
  di: Shen, Guobin, et al.
  Pubblicazione: (2025)
• Bidirectional Intention Inference Enhances LLMs' Defense Against Multi-Turn Jailbreak Attacks
  di: Tong, Haibo, et al.
  Pubblicazione: (2025)
• Jailbreaking Leaves a Trace: Understanding and Detecting Jailbreak Attacks from Internal Representations of Large Language Models
  di: Kadali, Sri Durga Sai Sowmya, et al.
  Pubblicazione: (2026)
• Efficient LLM Safety Evaluation through Multi-Agent Debate
  di: Lin, Dachuan, et al.
  Pubblicazione: (2025)