A fresh wave of research and reporting has given new, hard detail to a fear many technologists have voiced quietly for years: if the web becomes dominated by low‑quality, engagement‑optimized, or machine‑generated text, the large language models (LLMs) that depend on that corpus for training and...