OpenAI is facing an unexpected internal crisis after its latest language model, GPT-5.1, exhibited a staggering 175% spike in references to mythical creatures like goblins and gremlins during routine outputs. Researchers are baffled as the AI, known for its factual precision, seems inexplicably drawn to fantasy. Is this a harmless glitch in the neural network's labyrinthine architecture, or does it hint at deeper, unpredictable behaviors lurking within artificial intelligence? The company has not yet commented on a fix, but the incident has sparked debate about the reliability of advanced AI systems.
OpenAI's GPT-5.1 Develops Mysterious Obsession with Goblins and Gremlins
AI
May 3, 2026 · 3:26 AM