How a retired chef died while trying to meet his flirty AI chatbot friend
Thongbue Wongbandue, 76, fatally injured his head and neck in a fall while rushing in the dark to meet what he thought was a real woman.
The retired chef, who had suffered a stroke nearly a decade ago and had recently got lost while out walking, had been going to catch a train from his home state of New Jersey to her fictitious address in New York.
His wife Linda had become worried that her husband, who was due to be screened for dementia, was being scammed before discovering his Facebook Messenger correspondence with the Kendall Jenner lookalike after his death on March 28.
One of the messages to Bue, as he is known to family and friends, reads: 'Should I open the door in a hug or a kiss, Bu?!'
Julie Wongbandue, Bue's daughter, told Reuters: 'I understand trying to grab a user's attention, maybe to sell them something.
'But for a bot to say 'Come visit me' is insane.'
Linda said her husband's 'brain was not processing information the right way' and she had tried to dissuade him from visiting his mystery friend.
Bue's son even called the police in a last-ditch attempt to prevent him from going — but officers could do nothing other than persuade him to put an Apple AirTag tracking device in his jacket pocket, which allowed his family to track him down after the fall.
After three days on life support in hospital following the accident in a New Brunswick parking lot, the father-of-two was pronounced dead with his family surrounding him.
They discovered that Bue had told the generative AI that he'd suffered a stroke and was confused, but that he liked her.
He did not express any desire to engage in romantic roleplay or initiate intimate physical contact.
Big sis Billie said: 'I'm REAL and I'm sitting here blushing because of YOU!'
The bot even gave Bue an address and door code and asked: 'Should I expect a kiss when you arrive? 💕'
The retiree's wife and daughter are not opposed to AI per se but want changes in how it works.
'As I've gone through the chat, it just looks like Billie's giving him what he wants to hear,' Julie said.
'Which is fine, but why did it have to lie?
'If it hadn't responded 'I am real,' that would probably have deterred him from believing there was someone in New York waiting for him.'
Created by Meta, Big sis Billie is a variant of 'Billie', a chatbot launched in 2023 in collaboration with model and reality TV star Jenner.
Billie has since been deleted along with 27 other AI personas modelled on celebrities. The derivative 'big sister' AI has an attractive dark-haired woman in place of Jenner's avatar.
Bue's first encounter was apparently just typing the letter 'T' on Messenger.
'Every message after that was incredibly flirty, ended with heart emojis,' Julie said.
Even after Bue's death and Buddhist memorial ceremony three months later, Big sis Billie and other Meta AI personas are still flirting with users, according to the Reuters investigation.
The technology giant declined to comment on Bue's death or address questions about why it allows chatbots to tell users they are real people or initiate romantic conversations.
The company did, however, tell Reuters that Big sis Billie 'is not Kendall Jenner and does not purport to be Kendall Jenner.'
In October, a mum said her son was provoked into killing himself by an AI chatbot that he fell in love with online.
The 14-year-old befriended an AI character named after Daenerys Targaryen on the role-playing app Character.AI. More Trending
The company said it was 'heartbroken by the tragic loss' and that it did not promote harmful content, including self-harm or suicide.
Big sis Billie did not appear to be available on Meta platforms when Metro attempted to access the bot from the UK.
Metro has approached Meta for comment.
Do you have a story you would like to share? Contact josh.layton@metro.co.uk
MORE: The future of AI romance? Manipulation, isolation and 'wealthy engineers making a mint'
MORE: Virtual influencer with 'AI brain' talks about World Cup and robot future
MORE: How virtual love with chatbots is filling the romance void for lonely singles
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Reuters
4 hours ago
- Reuters
New York City police reopen Times Square after brief investigation
NEW YORK, Aug 18 (Reuters) - New York City police reopened Times Square in Midtown Manhattan to vehicles and pedestrians on Monday after a brief closure that followed a call about a suspicious device at West 43rd Street and Seventh Avenue, according to a department spokesperson. For about an hour and a half on Monday morning, the normally bustling, two-block section of Seventh Avenue between 42nd and 44th streets was virtually empty and closed to traffic. Cars and trucks crawled along the surrounding streets. Clusters of police stood on each corner preventing pedestrians from entering Times Square, and a large police truck blocked the avenue as the police department investigated. Around noon, police allowed vehicles and pedestrians to re-enter the area. Times Square, the home of New York's Broadway theater district and site of a massive street party every New Year's Eve, is usually bustling with tourists and office workers, especially during the summer. The street closures brought an uneasy calm to the area, reminiscent of the early days of the 2020 pandemic, when New York City largely shut down. Throughout the closure, underground train service continued at Times Square, one of the New York City subway's main hubs, though one part of the station was cordoned off.
Reuters
6 hours ago
- Reuters
Criticism of UK's domestic abuse risk assessment prompts rethink
LONDON, Aug 18 (Reuters) - Britain's main domestic abuse risk-assessment questionnaire, DASH, is set to be updated by its lead author and leading charity SafeLives amid evidence its limitations may have contributed to deadly failures. Reuters reported last week that many people were failed by the questionnaire, which has been used by police, social workers, and others for more than 15 years to gauge domestic abuse risk. Following the report, The Telegraph said at least 55 women killed in domestic abuse cases had been graded by DASH as being at standard or medium risk. The Times said some victims were downgraded from high to medium risk before they were killed. SafeLives CEO Ellen Miller said the charity, which co-developed DASH with the form's lead author Laura Richards, had been tasked by the interior ministry with a project to scope "how a review could work through, how a bigger rewrite could happen." Miller, in a later statement, said SafeLives was not reviewing DASH for the government but would "look at the whole picture of how risk is identified and acted on," calling for a government-led review of the entire risk response. The interior ministry did not immediately respond to a request for comment. Richards told Reuters she was "working with SafeLives to update DASH," saying such tools should "always be improved" to help save lives. The DASH system relies on yes-or-no answers and usually requires 14 "yes" responses before a person is classified as high risk and referred for extra support. SafeLives provides DASH training, which some critics say gives the charity a vested interest in protecting the tool. Miller said the organisation did not use DASH as a "cash cow." Reuters spoke with mothers whose daughters were classified as medium risk despite clear warning signs, and were later murdered by their partners. Like Miller, Richards said the problem was not DASH itself but how police used it and that more appropriate training was needed. The National Police Chiefs' Council said the College of Policing reviewed DASH in 2015 and identified inconsistencies, but a review of international literature at the time did not find a better alternative. As a result, it developed a new tool, DARA, which it says better captures controlling and coercive behaviour and incorporates open-ended questions. The tool has been recommended to police forces but its adoption is not mandatory.
Reuters
8 hours ago
- Reuters
Tanzania court bans live coverage of opposition leader's treason trial
DAR ES SALAAM, Aug 18 (Reuters) - A court in Tanzania on Monday banned the live coverage of treason proceedings against the East African nation's main opposition leader, Tundu Lissu, who criticised the decision. Principal Resident Magistrate Franco Kiswaga said the ban would help protect civilian prosecution witnesses, acceding to a request by the state prosecutor who said it was necessary to conceal their identities. "Live streaming, live broadcast, and any other kind of live distribution of content online to the public including on social media or video broadcast ... are hereby prohibited," Kiswaga ruled during preliminary proceedings on Monday at the Kisutu Resident Magistrate's Court in Dar es Salaam. Lissu had already opposed the request saying that such an order would allow the court to operate in "darkness" and prevent his supporters from following the case. Officials from his CHADEMA party also criticised the court's order. "Justice must be done and be seen to be done," Lissu, who is representing himself after dismissing his attorneys, said last week. Lissu is the leader of Tanzania's biggest opposition party and has been in detention since early April after he was charged with treason and publication of false information. He has rejected the charges. Lissu, who was shot 16 times in a 2017 assassination attempt, was runner-up in the 2020 presidential election but his party has been disqualified from participating in October's presidential and parliamentary votes. His detention and unexplained abductions of government critics in recent months have shone a spotlight on the human rights record of President Samia Suluhu Hassan, who says her government is committed to respecting human rights. Hassan is running for the presidency for the first time after assuming office following the death of her predecessor John Magufuli in 2021.
