A man in Shanghai, China, has been scammed out of almost $28,000 after becoming a victim of an AI-powered romance scam, according to reports from Chinese state media. The incident underscores a troubling trend: the fusion of artificial intelligence with traditional romance fraud, allowing criminals to craft incredibly convincing fake personas.

Scammers used generative AI software to produce lifelike videos and photographs of a young woman they called ‘Ms. Jiao.’ The victim, who became romantically interested in this fictional character, eventually transferred nearly 200,000 yuan (approximately $28,000) to a bank account controlled by the fraudsters. To strengthen the illusion, the scammers created a full fake identity and even fabricated medical records, using these documents to pressure the victim into believing the woman urgently needed financial help.

The Rise of AI-Driven Romance Fraud

Romance scams have existed for decades, but the emergence of generative AI has elevated them to a new level of sophistication. AI can now generate hyper-realistic facial images, voices, and even real-time video conversations. Tools like deepfakes enable scammers to produce convincing video calls or audio messages, making it increasingly difficult for victims to distinguish reality from fabrication.

In the Shanghai case, the scam began through a dating app or social media platform (exact details remain undisclosed). The victim received a friend request from an attractive profile, followed by frequent, emotionally charged conversations. Over time, the scammers built trust and then introduced a series of fabricated emergencies—medical crises, sudden bills, or urgent travel expenses. The victim, believing he was helping a real person in distress, sent multiple payments. Only when the requests became relentless and the victim attempted to verify the woman’s identity did he realize the deception.

Global Statistics and Alarming Trends

This incident is not isolated. Cybersecurity firm McAfee released a report on February 11, 2025, revealing that more than half (52%) of online daters have either been scammed out of money or pressured to send money or gifts to someone they met through digital platforms. The company observed an “explosion of online romance fraud” across social media, messaging apps, and AI chatbots. In their study, 26% of respondents said they or someone they know had been approached by an AI chatbot posing as a real person on a dating app or social media.

Additionally, 21% of those surveyed reported being contacted by someone pretending to be a well-known public figure. Among those who fell for it, 33% lost money, with an average reported loss of $1,985. The pre-Valentine’s Day period saw McAfee block 321,509 fraudulent URLs specifically designed to lure victims into romance scams. This indicates a concentrated effort by criminals to exploit emotional vulnerability during peak loneliness seasons.

Another high-profile case involved a French woman who was duped out of €830,000 ($850,000) after being convinced she was in a relationship with Hollywood star Brad Pitt—using AI-generated video and audio. That scam lasted over a year before the victim’s family intervened.

How AI Romance Scams Work

Modern AI romance scams typically follow a multi-stage blueprint:

• Profile Creation: Criminals use generative AI to create a unique, physically appealing avatar. They may employ GANs (Generative Adversarial Networks) to produce a person who doesn't exist, complete with a consistent facial features across images.
• Voice and Video Deepfakes: AI voice cloning tools, available cheaply online, allow scammers to make phone calls or send voice notes that sound genuine. For video, deepfake technology can synchronize a real person's facial movements with an script—sometimes by substituting the scammer's own face with the avatar's.
• Emotional Manipulation: Chatbots or human operators use AI-generated conversation scripts steeped in psychological triggers: urgency, sympathy, romance. They cultivate ‘love bombing’ then manufacture crises.
• Fake Documentation: Scammers forge medical records, hospital bills, or travel documents using AI image generators. In the Shanghai case, these documents were personalized with the victim’s name and the fictional woman’s details.
• Payment Pressure: Victims are urged to send money via wire transfers, gift cards, or cryptocurrency—methods that are hard to trace and recover.

The barrier to entry for such scams is low. Free or cheap AI tools are widely accessible. A scammer can generate a believable profile in minutes and then automate initial conversations using language models. Some operations now use AI to manage hundreds of simultaneous victims, dramatically scaling their reach.

Impact on Victims and Society

Financial losses from romance scams are staggering. In the United States alone, the Federal Trade Commission (FTC) reported losses exceeding $1.3 billion in 2023, with median individual losses around $4,400. AI-fueled scams are likely driving these numbers higher. Beyond money, victims suffer profound emotional trauma, including shame, depression, and eroded trust. Many are reluctant to report the crime due to embarrassment.

Law enforcement faces new challenges. Traditional identity verification becomes nearly useless when scammers can produce convincing government-issued ID facsimiles. Cross-border jurisdiction and the anonymity of cryptocurrency further complicate investigations. Some police forces are now training officers to recognize AI-generated evidence, but the pace of technological change outstrips resources.

Prevention and Awareness Strategies

Experts recommend several precautions to avoid falling prey to AI romance scams:

• Reverse Image Search: Use tools like Google Images or TinEye to check if profile pictures appear on other suspicious sites or known scam databases.
• Video Call Verification: Insist on live video calls. While AI deepfakes can be used, real-time calls are harder to fake perfectly. Look for glitches in lighting, mouth movements, or unnatural speech patterns.
• Be Wary of Excuses: Scammers often avoid meeting in person or video-calling by citing travel, illness, or technical issues. If someone resists video evidence, it’s a red flag.
• Guard Personal Information: Never share financial details, Social Security numbers, or passwords with someone you haven't met in person and verified.
• Verify Medical and Financial Stories: If a romantic interest claims to have a medical emergency, ask for specific hospital details and then independently contact that hospital. Legitimate institutions will have records.
• Use AI Detection Tools: Emerging software can analyze images and text for AI generation markers. While not foolproof, they can provide additional clues.
• Report Suspicious Activity: If you suspect a scam, cease communication and report the profile to the platform, local police, and national fraud agencies like the FTC or Action Fraud.

Technology companies are also stepping up. Meta, for example, uses machine learning to detect fake accounts that exhibit patterns of romance scam behavior. Dating apps like Tinder and Bumble are testing identity verification features. However, these efforts are reactive, and scammers continuously adapt.

The Psychological Underpinnings

Why do intelligent people fall for these scams? Romance scams exploit fundamental human needs for connection, intimacy, and validation. AI enhances the deception by relentlessly feeding the victim’s emotional hunger with personalized, timely responses. The victim may be lonely, recently widowed, or socially isolated. Scammers engineer a fantasy relationship that feels authentic, then weaponize the victim’s investment—both emotional and financial.

Cognitive biases also play a role. The sunk cost fallacy makes it hard for victims to walk away after they’ve already sent money; they hope to recoup losses or keep the relationship alive. Confirmation bias leads them to ignore warning signs and focus on affectionate messages. In the Shanghai case, the victim likely rationalized each payment as a necessary act of love, ignoring the growing inconsistencies.

Regulatory and Legal Responses

Governments worldwide are grappling with how to regulate AI-generated content to prevent misuse. China has already implemented strict deepfake laws requiring clear labeling of AI-generated media. However, enforcement across international borders remains weak. In the European Union, the AI Act classifies deepfake fraud as a high-risk application, requiring disclosure. The United States has introduced bills like the Deepfake Fraud Prevention Act, but none have become law yet.

Financial institutions are also adapting. Banks are training fraud detection systems to flag transactions that match patterns common to romance scams—multiple small payments to new accounts, sudden transfers to cryptocurrency exchanges, or withdrawals by elderly customers. Some institutions have started to delay transactions for customers who appear to be under emotional duress.

Future Outlook

As AI technology becomes cheaper and more sophisticated, romance scams will only grow more convincing. Real-time AI avatars that can hold conversations, adapt to user prompts, and mimic emotional cues are already being developed for customer service—but the same technology can be repurposed for fraud. The McAfee report predicts that 2025 will see a new wave of “hyper-personalized” scams where AI profiles learn victim preferences from social media and previous chats.

Meanwhile, detection methods are advancing. Researchers are building systems that analyze facial micro-expressions, voice stress patterns, and conversation pacing to identify deepfakes. Public education campaigns are increasing awareness, but the most effective defense remains skepticism and a willingness to verify.

The Shanghai victim’s story is a cautionary tale, but it is also a call to action for tech platforms, law enforcement, and individuals. In an age where anyone can be anyone online, trust must be earned through verifiable, offline interactions. As scammers harness AI to perfect illusions, the rest of us must sharpen our ability to spot the cracks.

Source: ReadWrite News