The Trigger in the Haystack: Extracting and Reconstructing LLM Backdoor Triggers

Documenti analoghi

• A Representation Engineering Perspective on the Effectiveness of Multi-Turn Jailbreaks
  di: Bullwinkel, Blake, et al.
  Pubblicazione: (2025)
• BadLLM-TG: A Backdoor Defender powered by LLM Trigger Generator
  di: Zhang, Ruyi, et al.
  Pubblicazione: (2026)
• PyRIT: A Framework for Security Risk Identification and Red Teaming in Generative AI System
  di: Munoz, Gary D. Lopez, et al.
  Pubblicazione: (2024)
• Revisiting Training-Inference Trigger Intensity in Backdoor Attacks
  di: Lin, Chenhao, et al.
  Pubblicazione: (2025)
• Invisible Textual Backdoor Attacks based on Dual-Trigger
  di: Hou, Yang, et al.
  Pubblicazione: (2024)