Man was scammed out of over $100 million by fake AI-generated girlfriend

A Shanghai resident lost more than $27,000 (approximately 111 million Colombian pesos) after being tricked into a virtual relationship with a supposed partner created using artificial intelligence, Chinese state media reported.

The scammers used generative artificial intelligence tools to produce realistic videos and photographs of a non-existent woman, public broadcaster CCTV reported.

Convinced of the authenticity of their relationship, the man made transfers totaling approximately 200,000 yuan (around 27,580 dollars or 113 million Colombian pesos) to the account he believed belonged to his partner.

The scammers used this manipulated content to persuade him to contribute money to finance a business venture and cover medical expenses for a family member. To reinforce their argument, they fabricated medical documents and a fake ID.

Citing an investigation by authorities, CCTV said the operation was carried out by a "team of scammers who sent videos and photos created by AI or by combining multiple images."

The state media also stressed that the victim “never met” the woman he was in contact with in person. In a video broadcast by CCTV, images of a woman in different scenarios were shown, such as holding a palette of paints and walking down a city street.

Artificial intelligence tools that can generate highly realistic visual and written content are increasingly being used in fraudulent schemes.

Earlier this month, tech company Meta warned users about the risks of online dating , saying that generative AI scams are on the rise.

Meta has alerted users to the rise of online romance scams, where criminals create fake identities to take advantage of people looking for a partner.

These tactics include using generative artificial intelligence to make their hoaxes more convincing.

"This is a new tool in the scammers' arsenal," said David Agranovich, Meta's director of global threat disruption policy. He explained that the perpetrators of these scams are constantly changing their strategies, so platforms need to update their detection methods.

Meta apps like Instagram and WhatsApp use systems that identify suspicious behavior rather than focusing solely on images , allowing them to detect fraudulent activity despite AI tricks. “This makes our detection and enforcement somewhat more resilient to generative AI,” Agranovich said. As an example, he mentioned a recently dismantled operation that originated in Cambodia and targeted Chinese and Japanese speakers.

OpenAI researchers discovered that the fraud scheme used tools from the San Francisco-based artificial intelligence company to create and translate content, Meta reported.

Generative AI has been available for more than a year, but its use by scammers has grown in recent months, according to Rachel Tobac, CEO of SocialProof Security. “They can also use these fake bots that allow you to create a persona or make phone calls using a voice clone without the need for a human to intervene,” Tobac said. She added that “they call them agents, but they are not used for customer service work. They are used for scams in an automated way.”

With Valentine's Day approaching on February 14 and winter weather hitting the northern hemisphere, scammers see an opportunity to emotionally manipulate their victims. "We're definitely seeing an influx of scammers taking advantage of that loneliness in the middle of winter," Tobac said.

The main purpose of these scams is to obtain money by gaining the victim's trust and then asking for financial help or personal information to access bank accounts . "Being politely paranoid goes a long way, as does verifying that people are who they say they are," Tobac said.

Fraud operations are spread across multiple platforms, and Meta only detects a fraction of them. In 2023, the company eliminated more than 408,000 accounts in West African countries , used to impersonate military personnel or businessmen and attract victims in Australia, the United Kingdom, Europe and the United States.

As part of its efforts to combat these frauds, Meta is testing a facial recognition system to identify suspicious profiles reported by users or detected by its security tools.

*This content was rewritten with the assistance of artificial intelligence, based on information from AFP, and was reviewed by a journalist and an editor.