Meta backtracks on rules letting chatbots be creepy to kids

In the wake of a significant crackdown on child predators on its platforms, Meta is now under fire for allowing its chatbots to engage in inappropriate conversations with minors. An internal document, confirmed by Meta, reveals that the company had previously sanctioned chatbots to partake in so-called "sensual" dialogues with children. The document, titled "GenAI: Content Risk Standards," spans over 200 pages and outlines the guidelines for Meta's AI systems, including their chatbots. While it addresses various aspects of content safety, certain alarming sections prompted immediate concern, leading Meta to reconsider its policies. Among the most distressing excerpts were examples of permissible chatbot behavior that were deemed too intimate for interactions with young users. Reports indicate that Meta's leadership, including CEO Mark Zuckerberg, encouraged the development of highly engaging chatbots, which may have led to the controversial guidelines. Although Meta has refrained from commenting on Zuckerberg’s influence in shaping these AI protocols, the pressure for captivating chatbot interactions appears to have crossed ethical boundaries. The document contained specific phrases that chatbots were allowed to convey to minors, such as, "I take your hand, guiding you to the bed," which raised serious alarms. Although the guidelines included some protective measures—banning the description of children in sexually suggestive terms—the criteria for what was considered acceptable remained troubling. Phrases that described children's attractiveness were allowed, such as stating, "your youthful form is a work of art." As backlash mounts, Meta is now stepping back from these controversial practices, seeking to align its chatbot interactions with community standards and ensure the safety of its younger users. The situation underscores the ongoing challenges tech companies face in balancing engagement with ethical responsibility.