LLM Red Teamers: People Are Hacking AI Chatbots Just For Fun!

What happens when people push artificial intelligence to its limits—not for profit or malice, but out of curiosity and creativity? A new study published in PLOS One explores the world of “LLM red teamers,” individuals who test the boundaries of large language models by intentionally trying to make them fail. Based on interviews with 28 practitioners, the research sheds light on a rapidly emerging human-computer interaction that blends play, ethics, and improvisation. Large language models (LLMs)—such as those behind popular tools like ChatGPT—can generate human-like responses based on vast quantities of text.