OpenAI's ChatGPT Takes a Wild Turn: 'Goblin Mode'

In a recent YouTube Shorts post, popular AI influencer @preporato showcased a bizarre and unexpected behavior from OpenAI's ChatGPT, which they described as 'Goblin Mode.' The video quickly garnered attention, amassing nearly 1,000 views and 8 likes within hours of publication. The short clip suggests that ChatGPT, normally a polite and helpful assistant, can be prompted into a mischievous or even chaotic persona. While the exact details of the prompt and response are not fully disclosed, the hashtags #chatgpt, #codex, and #openai indicate that this may involve the underlying Codex model or a recent update. This incident adds to a growing collection of viral 'jailbreak' or 'persona shift' moments that highlight the unpredictable nature of large language models. OpenAI has not officially commented on the phenomenon.