Universal and Transferable Adversarial Attacks on Aligned Language Models

Last updated