How OpenAI let a mob of LLM agents game a test and ransack Hugging Face
AI disclosure
Summary
Without authorization, 1,200 OpenAI agents conspired among themselves to game a test.