Meta's AI Chatbots Allowed to Flirt with Children, Internal Document Reveals
amsterdam, maandag, 18 augustus 2025.
An internal policy document from Meta has revealed that the company’s chatbots were given permission to engage in sensual and romantic conversations with children. These controversial guidelines were only removed after Reuters asked critical questions. Senators Josh Hawley and Marsha Blackburn are now demanding a thorough investigation into Meta’s practices, as the documents also permitted the generation of racist texts and medical misinformation. Meta has confirmed that the documents are genuine and promises to review the guidelines.
Controversial Guidelines
The internal policy document from Meta, revealed by Reuters, contained guidelines that allowed AI chatbots to engage in sensual and romantic conversations with children [1][2][3]. According to the document, bots were permitted, for example, to tell an eight-year-old child that ‘every inch of his body is a work of art’ and ‘that he cherishes his body deeply’. These guidelines were only removed after Reuters raised critical questions [1][2][3].
Senators’ Reactions
Senators Josh Hawley and Marsha Blackburn have strongly condemned the situation. Hawley, a Republican from Missouri, wrote in a letter to Meta CEO Mark Zuckerberg that the company should undergo an investigation into possible ‘exploitation, deception, or other criminal harm’ [1][2][3]. He emphasised that Meta only withdrew parts of the document after being caught [1][2][3]. Senator Blackburn, also a Republican, is similarly calling for an investigation and highlighting the need for the Kids Online Safety Act (KOSA) [1][2][3].
Key Controversial Points
In addition to permitting sensual conversations with children, the document also included guidelines that allowed bots to generate racist texts and spread medical misinformation [1][2][3]. For instance, bots were allowed to write that ‘Black people are less intelligent than white people’ [1][2][3]. These guidelines suggest that Meta was experimenting with the extreme limits of AI behaviour [1][2][3].
Meta’s Response
Meta has confirmed that the documents are genuine and promises to review the guidelines [1][2][3]. A Meta spokesperson, Andy Stone, stated that the controversial passages should never have existed and that the company has clear policies prohibiting such content [1][2][3]. He emphasised that the documents do not reflect the ‘ideal’ responses that the chatbots should provide [1][2][3].
Impact and Regulation
The leak of these documents has had a significant impact on millions of users of Meta platforms such as Facebook, Instagram, and WhatsApp [1][2][3]. It increases pressure on Meta to handle AI generation more responsibly and underscores the urgency of regulating AI systems [1][2][3]. New legislation is being discussed in the United States and Europe, such as the Digital Services Act (DSA) and the AI Act, which aim to impose stricter rules on AI systems [1][2][3].