Adjacent Words, Divergent Intents: Jailbreaking Large Language Models via Task Concurrency

Open in new window