The Trojan Knowledge: Bypassing Commercial LLM Guardrails via Harmless Prompt Weaving and Adaptive Tree Search

Documenti analoghi

• Do LLMs Really Forget? Evaluating Unlearning with Knowledge Correlation and Confidence Awareness
  di: Wei, Rongzhe, et al.
  Pubblicazione: (2025)
• One Turn Too Late: Response-Aware Defense Against Hidden Malicious Intent in Multi-Turn Dialogue
  di: Shen, Xinjie, et al.
  Pubblicazione: (2026)
• Graph Transductive Defense: a Two-Stage Defense for Graph Membership Inference Attacks
  di: Niu, Peizhi, et al.
  Pubblicazione: (2024)
• Federated Classification in Hyperbolic Spaces via Secure Aggregation of Convex Hulls
  di: Prakash, Saurav, et al.
  Pubblicazione: (2023)
• Differentially Private Graph Diffusion with Applications in Personalized PageRanks
  di: Wei, Rongzhe, et al.
  Pubblicazione: (2024)