By: Gary Marcus   |   The Atlantic   |   March 2023
Bad actors could seize on large language models to engineer falsehoods at unprecedented scale.