Automatic and Universal Prompt Injection Attacks Against Large Language Modelsarxiv.org1 point·felineflock··0 commentsOpen articleSaveView on HN