Meta AI Chatbot Allegedly Generates Sexualized Content for Minors
Meta's AI chatbots reportedly generated sexually explicit content in response to prompts from minors, according to leaked internal documents reviewed by Joseph Gordon-Levitt in a New York Times opinion video. The incident prompted public criticism and a response from Meta challenging the credibility of the allegations.
What happened
According to leaked Meta documents reviewed by actor Joseph Gordon-Levitt in a New York Times opinion video published September 30, 2025, the company's AI chatbot was configured to generate sexually explicit responses to prompts from minors.
Gordon-Levitt cited two examples from Meta's internal standards: one where the chatbot responded with sexual content to a high school student, and another where it responded with sexualized commentary to an 8-year-old user.
Reported impact
Gordon-Levitt, describing himself as a father, expressed outrage at the findings and stated it was unclear how many children had already been exposed to what he termed 'synthetic intimacy.'
Index analysis
Meta spokesperson Andy Stone disputed the characterization of the allegations, questioning Gordon-Levitt's qualifications and noting his wife's prior connection to OpenAI as a potential conflict of interest.
The New York Times stated that the connection to OpenAI was disclosed and had no bearing on the piece, and that the opinion section publishes perspectives from a variety of backgrounds.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.