Ex-OpenAI staffer claims frontier models are finetuned on your successful chats

burkov · x · 2026-09-11

Stanislav Burkov argues bluntly that only an idiot would believe LLM labs aren't finetuning their frontier models on billions of free customer conversations that ended in success — meaning users' successful chats are effectively free training data.

Original post →

More from Models

Models channel →