AIs can generate near-verbatim copies of novels from training data
oldnetguy Monday, February 23, 2026
Summary
Large language models can generate near-verbatim copies of novels from their training data, posing challenges for copyright and intellectual property protection. Researchers found that models can reproduce long passages of text with high fidelity, raising concerns about the potential misuse of these technologies.
42
23
Summary
arstechnica.com