404 Media has detailed 'Project Lily,' OpenAI's internal codename for a program that hires hundreds of contractors to read real ChatGPT prompts and rate the chatbot's replies. According to the report, contractors are often paid over $50 an hour through third-party staffing firms, and while accounts are anonymous, reviewers see entire conversations — including sensitive personal details that slip past automated filters. The dashboard shown to reviewers can include a summary of a user's stored memories, which 404 Media says can reveal location, profession or personal context. OpenAI would not tell the outlet where this is disclosed to users, pointing only to a help page after publication. The piece notes other major labs run similar human-review programs. If this matters to you, check your data controls.