#Jailbreak Attacks

Articles (1)

• 🇺🇸 Targeting Alignment: Extracting Safety Classifiers of Aligned LLMs — 19/02/2026 [USA]

  arXiv:2501.16534v5 Announce Type: replace-cross Abstract: Alignment in large language models (LLMs) is used to enforce guidelines such as safety. Yet, alignment fails in the face of jailbreak attacks...

  Related: #Large Language Models, #Model Alignment, #Safety Classifiers, #LLM Security