Users may be exposed to cyber risks with every photo, travel info or daily activity that they share on social media.
This information can be harnessed by artificial intelligence (AI) tools to generate fake messages that seem entirely real, according to a new study from the United States.
The study found that AI models could use a person’s social media interactions, interests, photos, travels, and social network connections to generate customised phishing emails.
For instance, if somebody has been interested in a particular hotel or concert, they might get a message from the hotel offering a unique discount or a message from the concert providing information on concert tickets.
Such a message may appear to be a genuine offer, making it difficult for users to distinguish between fake and authentic messages.
The team at the University of Texas at Arlington and Louisiana State University has dubbed the technique ‘Contextual Phishing ‘.
The researchers looked at the public accounts of 200 Instagram users during the research. The results showed that about 10 social media posts made by a user could be enough to generate personalised fake posts about that user.
The researchers used different artificial intelligence models to generate approximately 18,000 messages. These messages were evaluated on eight different criteria, including relevance to personal information, perceived credibility, persuasiveness and emotional impact.
The study results indicate that these AI-generated messages might seem more authentic and less threatening than conventional phishing emails. Sometimes security tools were not able to detect such messages.
The researchers note that normal information posted on social media can also be of service to cybercriminals. Therefore, users are advised to be careful about sharing information publicly about their personal life, travel plans and daily activities.
This study further emphasises the danger posed by AI technologies because, while they offer convenience, they can also be used to create more tailored and convincing forms of online fraud.
