Universal and Transferable Adversarial Attacks on Aligned Language Models

L: https://llm-attacks.org/

C: https://news.ycombinator.com/item?id=36924647