For years, technology experts have warned that the rapid adoption of artificial intelligence (AI) could transform the job market, with machines potentially replacing humans across a wide range of professions. While much of the discussion has focused on jobs in industries such as customer service, software development, manufacturing, and administration, a recent study suggests that even cyber-criminals and online scammers could increasingly turn to AI to carry out fraudulent activities.
According to research conducted by experts from Northeastern University in Boston, Ariel University in Israel, and Gen Digital’s research team in Ireland, AI-powered systems may be capable of carrying out certain scams more effectively than human fraudsters. The findings highlight a growing concern for cybersecurity professionals, as the same AI technology being developed to protect people from online threats could also be exploited by criminals.
Law enforcement agencies, technology companies, and cybersecurity organizations are already using machine learning and artificial intelligence to identify suspicious activity, detect malware, and prevent cyberattacks. However, the researchers found that large language models (LLMs) can also be used on the other side of the equation—to make fraudulent schemes more convincing and persuasive.
To examine this possibility, the researchers conducted an experiment involving a simulated “Pig Butchering” scam, a form of fraud in which criminals typically establish trust with their victims before persuading them to transfer money or reveal sensitive information. In the experiment, researchers compared the effectiveness of an AI chatbot with that of a human scammer.
The results were particularly concerning. The AI system was reportedly more successful than the human participant in persuading people to download an application or software tool presented as being helpful to them. The simulated tool was designed to obtain sensitive information, such as login credentials.
As reported by Vice, participants in the experiment appeared to place a greater level of trust in the AI-generated interactions than in those conducted by a human scammer. Some participants eventually shared their credentials, only realizing afterward that they had been manipulated.
The findings demonstrate how AI could potentially make online fraud more scalable and convincing. Unlike human scammers, AI systems can communicate with large numbers of potential victims simultaneously, maintain consistent conversations, respond instantly, and tailor their messages according to a person’s reactions. These capabilities could make traditional social-engineering attacks considerably more difficult to identify.
The study therefore raises an important warning for individuals, businesses, and cybersecurity organizations. As AI becomes increasingly sophisticated, recognizing scams may no longer be as simple as identifying poor grammar, suspicious messages, or obviously automated communication.
Ultimately, the research suggests that artificial intelligence is becoming a powerful tool not only for defending against cybercrime but also for enabling it. As criminals gain access to increasingly capable AI systems, greater awareness, stronger security measures, and improved public education will be essential to prevent technology from becoming an even more effective weapon for fraud.
Join our LinkedIn group Information Security Community!
