⚡️ Hundreds of contractors are getting paid $50/hour to read real ChatGPT conversations

Internal documents call it "Project Lily." Reviewers read whole conversations between users and the bot, then rate the replies on a 1-to-7 scale 💀

They don't see usernames, and a filter model is supposed to strip out personal details first, but OpenAI's own documentation admits it misses things. Some tasks also come with a "user memories summary" attached, which can show what you've used the bot for before and roughly where you live 📍

This only applies if "improve the model for everyone" is on. It's on by default. Turning it off only covers new chats, not old ones ⚙️

Anthropic confirmed it does human review too, for accounts with the equivalent setting enabled.

🤑 @Dices the #1 Non-KYC Telegram Casino

😀 Join @CLASSIFIED