Hoxhunt blog

ChatGPT vs. human phishing and social engineering study: Who's better?

In the following study, we compared win rates on simulated phishing attacks between human social engineers and AI large language models. Pyry Åvist, Co-Founder and CTO of Hoxhunt, led the experiment with a 53,127-user-sample on the effectiveness of ChatGPT-generated phishing attacks. Many will find the study’s results both surprising and soothing. For the time being, human-generated phishing attacks remain the greater threat, but it’s a fluid situation. More importantly, good security awareness and phishing training protects email users against both human and AI-generated threats. You don’t need to reconfigure your security training to address the ChatGPT phishing threat just this minute. But having a good behavior change program in place is highly recommended.