UTSA researchers recently completed one of the most comprehensive studies to date on the risks of using AI models to develop software. In a new paper, they demonstrate how a specific type of error could …