Automatic and Universal Prompt Injection Attacks against Large Language Models

Open in new window