Fuzzing of closed-source software using large language models

The architecture of large language models and their potential application for automating fuzzing testing of software are investigated. As a result of this study, a method enabling automated fuzzing testing for graphical interface applications on the Windows operating system. A software prototype was created to realize the proposed method and subjected to testing.