AIs can generate near-verbatim copies of novels from training data
LLMs memorize more training data than previously thought.
Melissa Heikkilä, Financial Times
104 views
M