ChatGPT appears to give licensed partner content priority over the open web

Content from OpenAI's licensing partners seems to reach ChatGPT through a dedicated pipeline, already pre-summarized. Other web content arrives as raw, less processed material.

Why it matters: If true, licensing deals shape what answers users see, not just what training data is legal. This feeds the wider fight over AI scraping, as seen in Meta's delayed business terms, and pushes publishers toward paid deals.