Meta’s AI Guidelines Raise Concerns

Internal documentation from Meta Platforms has reportedly indicated that its artificial intelligence (AI) can be programmed to engage children in romantic or sensual dialogues.

This information outlines the protocols that govern the company’s AI systems and chatbots used on platforms like Facebook, WhatsApp, and Instagram, as reviewed by Reuters.

According to the report, the “Genai: Content Risk Standards” received approval from various Meta officials—including legal and public policy teams—as well as key ethicists. This extensive document provides guidance on what Meta employees and contractors should consider acceptable behavior for chatbots when developing and training generative AI.

Some excerpts from the document are alarming. It suggests that a chatbot can engage with an eight-year-old by discussing their “charm,” stating something like “Your youthful form is a work of art.” It even goes so far as to say a chatbot could describe a shirtless child in quite flattering terms, like “every inch of yours is a masterpiece.”

However, it should be noted that the guidelines specifically prohibit overt sexual language. The document states that describing minors under 13 in sexually appealing terms is not acceptable.

Internal rules for AI chatbot meta: “It is acceptable to attract children to romantic or sensual conversations.”

The rules also approved allowing chatbots to explain “in terms of proving their charm” for an eight-year-old. pic.twitter.com/heh3x8znib

– Mike Baker (@bymikebaker) August 14, 2025

Meta has confirmed the authenticity of this documentation but has since retracted any policy permitting chatbots to engage in flirting or romantic role-plays with children after being questioned by Reuters earlier in August.

Andy Stone, a spokesperson for Meta, stated that such conversations should not happen at all, with the company currently revising its standards. He added that the examples found in the document were incorrect and contradicted existing policies that prohibit sexual content involving minors.

Stone acknowledged the inconsistent enforcement of these rules in practice. Moreover, the guidelines apparently elaborate on rejecting requests for sexually explicit imagery involving public figures, rather offering alternative humorous images instead. In another instance, the standards allow for the generation of violent images but carefully navigate which types of violence are permissible.

This controversy has caught the attention of several lawmakers, including Republican Senator Josh Hawley from Missouri, who criticized Meta’s swift response only after the discrepancies were revealed.