ChatGPT Suffers Temporary Glitch, Displays Bizarre Responses Before OpenAI Rolls Out Fix
• ChatGPT began exhibiting bizarre, nonsensical responses on Tuesday, including repetitive loops and hallucinated content
• The issues became widespread, prompting OpenAI to investigate and identify a technical bug
• OpenAI rolled out a fix on Wednesday morning and confirmed ChatGPT was back to normal operation
• The incident highlights how generative AI models can unexpectedly change behavior
• OpenAI said the error occurred in how the model processes languages, leading to incorrect token mappings