AI Tools That Pinpoint Holiday Photo Locations with Over 90% Accuracy Are Being Weaponised in a New Class of Geolocation Fraud

Research published by McAfee, the cybersecurity company, has demonstrated that freely available AI models can accurately identify the location where a travel photograph was taken in more than 90% of cases, even without geotag data or attached metadata. Testing across more than 21,000 travel images found that one AI model achieved 91% accuracy and a second achieved 87%, identifying locations from contextual visual cues including architecture, signage, street markings, food stalls, and local light conditions. The research, reported on 16 August 2026, highlights a growing class of fraud in which criminals use AI geolocation analysis of publicly posted social media photographs to construct convincing phishing messages. The technique works by identifying where a target has travelled, then sending a text or email appearing to come from their bank or card provider, citing the specific location as evidence of suspicious activity. Because the victim has not disclosed their destination explicitly, the location-specific reference gives the message false credibility and increases the likelihood of the target clicking through and providing financial details. Vonny Gamot, head of EMEA at McAfee, described the mechanism as AI giving context to scam attempts, making them credible. The company found that even images with no recognisable landmarks could identify the country of origin, which is sufficient information for fraudsters to construct a targeted message. McAfee recommends delaying social media posts until after travel is complete and adjusting privacy settings to limit audience reach.