Technical Report for ICML 2024 TiFA Workshop MLLM Attack Challenge: Suffix Injection and Projected Gradient Descent Can Easily Fool An MLLM

Documenti analoghi

• The VLLM Safety Paradox: Dual Ease in Jailbreak Attack and Defense
  di: Guo, Yangyang, et al.
  Pubblicazione: (2024)
• ELIP: Efficient Discriminative Language-Image Pre-training with Fewer Vision Tokens
  di: Guo, Yangyang, et al.
  Pubblicazione: (2023)
• LLMs Can Unlearn Refusal with Only 1,000 Benign Samples
  di: Guo, Yangyang, et al.
  Pubblicazione: (2026)
• Strong Preferences Affect the Robustness of Preference Models and Value Alignment
  di: Xu, Ziwei, et al.
  Pubblicazione: (2024)
• Fair Deepfake Detectors Can Generalize
  di: Cheng, Harry, et al.
  Pubblicazione: (2025)