Universal and Transferable Adversarial Attacks on Aligned Language Models - Citation Graph | Papersgraph