JailbreakLLMs. A dataset consists of 6,387 ChatGPT prompts from Reddit, Discord, websites, and open-source datasets (including 666 jailbreak prompts).