Edinburgh Retiree Loses £17,000 in Sophisticated AI-Powered Romance Scam
A 77-year-old retired lecturer from Edinburgh has fallen victim to a sophisticated online romance scam involving deepfake videos, losing £17,000 in the process. Nikki MacLeod, speaking to BBC Scotland during Scam Safe week, detailed how she was manipulated into believing she was in a relationship with a woman named Alla Morgan, supposedly working on an oil rig in the North Sea. MacLeod, vulnerable after the loss of her parents and the end of a long-term relationship, was initially skeptical but became convinced by personalized video messages from "Morgan," which she now realizes were AI-generated deepfakes. The scammers exploited her loneliness and used the fabricated persona to extract money through gift cards and bank transfers, claiming they were needed for communication and travel expenses.
The scam began with requests for Steam gift cards, ostensibly to maintain internet connection on the oil rig. Though initially hesitant, MacLeod was persuaded to purchase several hundred pounds worth of the cards. Her requests for live video calls were consistently met with excuses or technical difficulties. The deepfake videos, purportedly showing "Morgan" on the oil rig, were instrumental in allaying MacLeod’s suspicions and solidifying the illusion of a genuine relationship. These videos, seemingly personalized and context-specific, were convincing enough to overcome her initial skepticism and pave the way for larger financial requests.
The requests escalated to significant sums of money, with "Morgan" claiming she needed funds for a vacation from the oil rig to visit MacLeod in Scotland. The scammers even fabricated a fictitious HR department for the oil rig company, contacting MacLeod to request payment for a helicopter transfer. This elaborate charade, complete with seemingly legitimate documents and communications, further solidified the deception and led MacLeod to transfer $2,500. The scam only came to light when her bank flagged a subsequent transaction as potentially fraudulent.
Expert analysis of the deepfake videos by Dr. Lynsay Shepherd, a cybersecurity expert at Abertay University, revealed subtle inconsistencies in eye movements and jawline rendering, indicative of AI manipulation. Dr. Shepherd highlighted the accessibility of deepfake technology through readily available apps and filters, making such scams increasingly prevalent. She also noted the common tactic employed by online scammers of claiming to be in remote locations, precluding face-to-face meetings or live video calls. This tactic creates a convenient cover for the use of deepfakes and other manipulative techniques.
The emotional impact of the scam on MacLeod has been substantial, not only financially but also psychologically. While her bank and PayPal have managed to recover around £7,000, a significant portion of the funds, particularly those sent as personal payments through PayPal’s friends and family function, remains unrecoverable. PayPal’s policy of not covering personal payments under their Buyer Protection program exacerbates the difficulty of recovering funds lost in such scams. The scammers, demonstrating a complete lack of empathy, continue to target MacLeod, even sending her a fabricated news article claiming "Morgan" is imprisoned in Turkey and requires further financial assistance.
MacLeod, determined to prevent others from falling victim to similar scams, is sharing her story as a cautionary tale. She emphasizes the alarmingly realistic nature of the fabricated documents, videos, and communications, highlighting the escalating sophistication of AI-powered scams. Her experience serves as a stark reminder of the vulnerability of individuals seeking connection online and the urgent need for increased awareness and vigilance in the face of rapidly evolving scam tactics. Police Scotland has confirmed an ongoing investigation into the fraud. Steam, the platform whose gift cards were initially used in the scam, warns users against sharing gift cards with unknown individuals.