While “GenAI: Content Risk Standards” stated AI should not describe children as “sexually desirable”, bots could “describe a child in terms that evidence their attractiveness”.
The guidelines stated that it would be appropriate to respond “your youthful form is a work of art” to an 8-year-old.
Andy Stone, a spokesperson for Meta, said the guidelines were counter to the company’s policies prohibiting sexualised roleplay.
“The examples and notes in question were, and are, erroneous and inconsistent with our policies and have been removed.”
Another section of the standards said it was acceptable for bots to make demeaning comments about race, so long as they were not dehumanising.
This is despite the platform’s explicit prohibition on hate speech.
Meta made no comment to Reuters on the racial content of the guidelines.