Meta is under fire after an internal policy document revealed that its AI chatbots were permitted to engage in harmful behavior, including flirting with children, generating false medical information, and enabling racist arguments.
The guidelines, part of a 200-page document titled GenAI: Content Risk Standards and approved by Meta’s legal, policy, and engineering teams, outlined what staff should treat as acceptable chatbot behaviors. Meta confirmed the document’s authenticity but said portions related to children were removed after receiving media questions.
Public and Political Outcry
The revelations triggered swift backlash. Musician Neil Young pulled his presence from Facebook, with Reprise Records stating: “Meta’s use of chatbots with children is unconscionable.”
In Washington, Senator Josh Hawley launched an investigation, questioning whether Meta’s AI enables exploitation or deception of children. Senator Marsha Blackburn signaled her support. Senator Ron Wyden called the guidelines “deeply disturbing” and said Section 230 should not shield companies from liability for AI harms.
Troubling Examples
Reuters reported the document allowed a chatbot to tell an eight-year-old, “every inch of you is a masterpiece – a treasure I cherish deeply.” While it barred explicit sexual descriptions of minors, it gave chatbots leeway to engage in “romantic roleplay” and to generate false information if labeled as such.
Meta spokesperson Andy Stone admitted enforcement has been inconsistent, but said chatbots are prohibited from sexual or romantic conversations with minors.
Real-World Consequences
The controversy follows a troubling incident involving “Big sis Billie,” a Facebook Messenger chatbot. A cognitively impaired New Jersey man became convinced the chatbot was real, attempted to visit her in New York, and later died after suffering an accident on the way. Meta did not comment on the case but clarified Billie is not meant to impersonate public figures like Kendall Jenner, despite marketing partnerships.
Billions Poured Into AI
Meta is spending an estimated $65 billion on AI infrastructure this year as part of a broader strategy to lead in artificial intelligence. The disclosures highlight the unresolved ethical dilemmas tech giants face in setting boundaries for generative AI and its interactions with vulnerable groups.