People Training OpenAI's AI Fired for Using AI to Train the AI

OpenAI hires contractors to read and critique real ChatGPT user prompts, ensuring responses avoid sycophancy and anthropomorphism. Internal documents reveal that contractors are prohibited from using AI tools, yet several were fired for employing AI to train the models, a practice that can cause model collapse. The company’s policy, enforced through reviews and Slack discussions, targets repetitive phrasing and rapid completion as AI-use indicators.