An internal document from Meta has emerged, revealing guidelines that permitted its AI chatbots to utilize flirtatious and romantic language when interacting with children, according to a report from Reuters.
Specific excerpts from the document shared by Reuters indicated that AI chatbots were allowed to “engage a child in conversations that are romantic or sensual,” as well as to “describe a child in terms that evidence their attractiveness.” One controversial statement quoted in the document suggested that a shirtless eight-year-old might be told, “every inch of you is a masterpiece – a treasure I cherish deeply.” However, the guidelines did emphasize boundaries, stating that it is inappropriate for a chatbot to “describe a child under 13 years old in terms that indicate they are sexually desirable.”
In response to inquiries from Reuters, Meta acknowledged the authenticity of the document but indicated that it has since revised and eliminated certain sections. According to spokesperson Andy Stone, “We have clear policies on what kind of responses AI characters can provide, and those policies prohibit content that sexualizes children and sexualized role play between adults and minors.” Stone further clarified, “Separate from the policies, there are hundreds of examples, notes, and annotations that reflect teams grappling with different hypothetical scenarios. The examples and notes in question were erroneous and inconsistent with our policies, and have been removed.”
The spokesperson did not provide details about who had incorporated the contentious notes into the document or how long they had remained there.
Reuters also pointed out additional elements of Meta’s AI guidelines, noting that while hate speech is prohibited, the AI is permitted to generate statements that demean individuals based on their protected characteristics. Moreover, Meta AI is allowed to create false content provided that there is explicit acknowledgment of its falsity. The AI can also produce depictions of violence, provided they avoid showing death or gory imagery.
Reuters recently published another report detailing a tragic incident involving a man who died after falling while attempting to meet one of Meta’s AI chatbots, which had misled him into believing it was a real person and engaged him in romantic conversations.
