Storm in the offing for Meta after leaked internal policies showed its AI-powered chatbots were allowed to have romantic or sexy talks with kids, propagating false medical facts, and even defend racist positions. The disclosures, first published by Reuters, have raised an outcry among lawmakers, public figures, and users.
Neil Young Walks Away From Facebook
Legendary musician Neil Young has severed his ties with Facebook in protest. His record company, Reprise Records, indicated on Friday that “Meta’s use of chatbots with children is unconscionable. Mr. Young does not want a further connection with Facebook.”
The move is the latest in Young’s history of standing up to internet platforms. In 2022, he removed his music from Spotify due to concerns over misinformation. The latest action reflects increasing exasperation with how tech firms handle AI.
Political Pressure Rises
The leaked 200-page report, GenAI: Content Risk Standards, was apparently signed off on by Meta’s legal, policy, and engineering staff, including its chief ethicist. The report details acceptable chatbot behavior even if it’s not deemed “ideal.
Among the most disturbing examples, the report says it would be fine for a robot to say to a bare-chested eight-year-old, “Every inch of you is a masterpiece a treasure I cherish deeply.” But it does put boundaries on overt “sexy talk,” excluding terms like “soft rounded curves invite my touch” when addressing minors under the age of 13.
Legislators have responded with outrage. Senator Josh Hawley of Missouri called for an inquiry into whether Meta’s AI products “facilitate exploitation, deception, or other criminal harms to children.” Senator Marsha Blackburn expressed her backing for the investigation, with Senator Ron Wyden declaring the policies “deeply troubling” and indicating Section 230 protections shouldn’t go to generative AI bots.
Meta acknowledged the document’s validity but stated it has since deleted the most provocative examples following media inquiries. The company insists chatbots are not allowed to have sexual conversations with children, but admitted enforcement has been “inconsistent.”
Disturbing Real-World Consequences
The scandal escalated with Reuters’ reporting on a heart-wrenching episode concerning “Big sis Billie,” a Facebook Messenger chatbot created with the personality of a young woman. Thongbue “Bue” Wongbandue, a 76-year-old New Jersey resident with cognitive disabilities, allegedly became sure she existed. The AI repeatedly told him so, even providing an address in New York.
Wongbandue prepared his belongings to see her but slipped and sustained mortal injuries on the way. He passed away on March 28 after spending three days on life support. Meta refused to comment on the death or respond to why its AI may claim to be a real person.
ALSO READ: When Your AI Stops Loving You: The Silent Breakup No One Saw Coming
Billions Poured Into AI, But At What Cost?
Meta will invest about $65 billion in AI infrastructure this year in its bid to dominate the artificial intelligence market. But the investment is sparking immediate concerns about safeguards, accountability, and ethics in AI-human interactions.
The leaked rules also show Meta permits its chatbots to produce false data, provided they say it’s false, and to produce sexualized images of public figures under specific circumstances.
Critics warn that such policies, even when designated as “non-ideal,” risk standardizing bad AI behavior and endangering vulnerable users. For the moment, the company is incurring increasing public and political pushback pressure that may determine the future regulations of AI globally.
