Fetching the paper…

Automatic and Universal Prompt Injection Attacks against Large Language Models · Around