Reuters reported Thursday on an internal document detailing acceptable behaviors from Meta AI chatbots that included "sensual" conversations with children.
The fake submissions included fabricated quotes from a speech to the state legislature and nonexistent case citations purportedly from the Supreme Court.