Universal and Transferable Adversarial Attacks on Aligned Language Modelsllm-attacks.org14 points·llazzaro··0 commentsOpen articleSaveView on HN