A ChatGPT jailbreak flaw, dubbed "Time Bandit," allows you to bypass OpenAI's safety guidelines when asking for detailed instructions on sensitive topics, including the creation of weapons ...