A Guideline for the Adoption of Generative AI to Support Secure Software Development Life Cycle (SSDLC): A Case Study of ChatGPT

Secure software development using generative AI is gaining traction, particularly in academic environments where undergraduate teams often lack structured guidance. A key challenge is aligning documentation outputs with formal standards such as OWASP ASVS when using large language models (LLMs) like ChatGPT-4o. This study proposes a Secure SDLC guideline incorporating a six-element prompting framework Context, Task, Instruction, Clarify, Refine, and Warning combined with OWASP ASVS principles. A case study involving a student-developed village information system applied the guideline from Project Charter to Secure Design phases. Results demonstrated improved completeness and standard alignment in early phases, though limitations were found in traceability during Business Requirements and Threat Modeling.