ChatGPT boosts software security by identifying vulnerabilities with high accuracy.
The researchers tested how well a big language model called ChatGPT could help students find and fix security issues in a web app. They gave ChatGPT some buggy code and found that it could spot most of the problems and even suggest ways to fix them. ChatGPT also found some extra issues on its own. Overall, ChatGPT did a good job at helping with security tasks in the software security course.