Benchmarking and Defending Against Indirect Prompt Injection Attacks on Large Language Models

Open in new window