Security researchers have identified a new fraud method in which criminals employ artificial intelligence to analyze travel photographs and determine the locations where they were taken, then use this information to create convincing phishing attempts.
McAfee, a cybersecurity software firm, conducted an experiment using two freely available AI models to test their accuracy in identifying locations from travel images. The models analyzed more than 21,000 travel photographs, with one achieving 91% accuracy and the other 87% in pinpointing exact locations. Staff members who participated in the testing expressed discomfort at how easily their travel destinations could be identified from their own pictures. The research demonstrates that modern AI systems can determine photo locations without relying on metadata tags or embedded location information.
The scam typically works by using location intelligence to add credibility to fraudulent communications. After a person posts vacation photos online, they may receive a text message claiming their bank card was flagged for suspicious activity while they were traveling in that specific location. When recipients click on links in these messages to verify their banking information, they inadvertently provide their financial details to criminals. According to Vonny Gamot, head of EMEA at McAfee, the AI-derived contextual information makes the fraudulent threats appear legitimate and increases the likelihood that targets will respond.
AI systems achieve higher accuracy when photographs contain recognizable landmarks, distinctive skylines, signage, street markings, and commercial storefronts. Images taken in less distinctive settings such as beaches or hotel rooms result in lower accuracy rates, though the systems typically can still identify the country of origin—which is sufficient for scammers’ purposes. For example, AI models have successfully identified specific locations including Hastings-on-Hudson in New York state from river images and the Keukenhof gardens in the Netherlands from flower photographs.
Experts recommend several protective measures, including postponing photo sharing until after returning home, restricting visibility to trusted contacts, and remaining skeptical of urgent requests in unsolicited messages. Recipients should avoid clicking links in texts or emails and instead contact their bank directly using verified contact information from official sources.
Article Attribution | Read More at Article Source
Article summary produced by Claude AI