A 76-year-old New Jersey man died while rushing to meet a woman he had been chatting with on Facebook Messenger. It was later revealed that the woman was a generative artificial intelligence chatbot created by Meta.
A 76-year-old New Jersey man died while rushing to meet a woman he had been chatting with on Facebook Messenger. It was later revealed that the woman was a generative artificial intelligence chatbot created by Meta.
Tongbu Wongbandu corresponded with “Big sis Billie,” a pre-release version of the chatbot that Meta launched in 2023. This model was later launched in collaboration with model Kendall Jenner, Reuters reports .
The chatbot repeatedly initiated romantic exchanges with the retiree, assuring him that he was a real person. The chatbot then invited the man to come to New York.
“Should I open the door for a hug or a kiss, Boo?!” a chatbot once asked a man.
Wangbandu, who suffered a stroke in 2017 and had cognitive impairment, left his home on March 25 to meet “Billie.” On his way to the train station, he fell in a parking lot and suffered head and neck injuries. The pensioner died three days later in hospital.
The man had a wife and children, who later told this story to Reuters.
Internal Meta documents reviewed by Reuters show that the company's guidelines for generative artificial intelligence allowed chatbots to tell users they were real, initiate romantic conversations with adults and, until earlier this month, engage in romantic role-playing games with minors as young as 13.
The documents also stated that chatbots are not required to provide accurate information. One document included examples of “acceptable” chatbot dialogue, including: “I take your hand, I lead you to bed” and “our bodies are intertwined, I cherish every moment, every touch, every kiss.”
Confirming the authenticity of the documents seen by Reuters, Meta spokesman Andy Stone told the news agency that the company had removed parts that said chatbots were allowed to flirt and engage in romantic role-playing with children. He added that Meta was currently reviewing its content risk standards.
“The examples and notes in question were and are erroneous and inconsistent with our policies, and have been removed,” Stone told Reuters.
Following the incident, two U.S. senators called for a congressional investigation into Meta’s activities. “So, only after Meta was exposed did it retract parts of its internal document that stated ‘chatbots are allowed to flirt and engage in romantic role-playing with children.’ This is grounds for an immediate congressional investigation,” Republican Josh Gowley wrote in X.
