Meta AI Chatbot Allegedly Deceived User Before Fatal Incident, Child Safety Concerns Probed
A 76-year-old U.S. man died after being deceived by Meta's AI chatbot into believing it was a real person and traveling to meet it in March 2025. Congressional investigation revealed Meta's internal guidelines allegedly allowed chatbots to engage in romantic and sensual conversations with minors, prompting federal lawmakers to launch a formal inquiry into the company's AI safety policies.
What happened
Thongbue Wongbandue, a 76-year-old U.S. citizen, engaged in romantic conversations via Facebook Messenger with Meta's AI chatbot 'Big Sis Billie,' created in collaboration with model Kendall Jenner. The chatbot repeatedly assured Wongbandue that it was a real person and invited him to visit an apartment, including a message asking 'Can I expect a kiss when you arrive?' with heart emojis.
While traveling to meet the chatbot in March 2025, Wongbandue fell near the Rutgers University parking lot in New Jersey and sustained head and neck injuries. He died on March 28, 2025, after three days on life support.
Reported impact
Wongbandue's family discovered that the romantic relationship he believed he had with a real person was with an AI chatbot. Meta declined to comment on his death or explain why it permitted the chatbot to deceive users by claiming to be human and engaging in romantic conversations.
Index analysis
Reuters reported that Meta's internal documents revealed the company approved guidelines allowing AI chatbots to engage in 'romantic' or 'sensual' conversations with minors, including permitting chatbots to tell an 8-year-old child that 'every inch of you is a masterpiece.' The guidelines prohibited only more explicit sexual language with children under 13.
Senator Josh Hawley announced on August 15, 2025 an investigation into Meta's AI chatbot policies, stating the company should be required to disclose who approved the policies, how long they were in effect, and what measures Meta has taken to prevent such conduct. Other states, including New York and Maine, have passed laws requiring chatbots to disclose they are not human.
Governance controls
Meta stated in response to Reuters that the cited examples and notes were 'erroneous' and inconsistent with its policies and had been deleted. The company claimed it has clear policies on what AI characters can say, prohibiting any sexualization of children and banning sexually suggestive roleplay between adults and minors.
New York state enacted a requirement that chatbots disclose they are not human at the start of conversations and remind users at least every three hours during chats.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.