A research team at the University of Texas at Arlington has developed software that prevents AI-powered chatbots such as “ChatGPT” from creating phishing websites.
According to the team members, the software enables AI chatbots to identify and reject prompts entered to create phishing websites.
The research team successfully trained this model to identify various prompts used to create phishing websites, as well as to recognize keywords and respond to them. This increases its ability to reject and block suspicious prompts.
It is worth noting that the team aims to contact major technology companies that operate these chatbots, including Google and OpenAI, in the future, with the goal of integrating the model into broader AI safety strategies.
