📄 Abstract
Artificial Intelligence (AI) has transformed software development by enabling automated code generation through advanced Large Language Models (LLMs) such as ChatGPT, GitHub Copilot, and Amazon Code Whisperer. These intelligent systems assist developers by generating source code quickly and efficiently, reducing development time and improving productivity. However, the widespread use of AI-generated code has raised concerns regarding code originality, academic integrity, software ownership, plagiarism, and cybersecurity. Distinguishing AI-generated code from human-written code has therefore become a significant research challenge. This research presents a Machine Learning-based approach for detecting AI-generated source code. The proposed system employs Natural Language Processing (NLP) preprocessing techniques, TF-IDF (Term Frequency–Inverse Document Frequency) feature extraction, and a Support Vector Machine (SVM) classifier to identify whether submitted code is written by a human programmer or generated by an AI system. A labelled dataset containing Human-written and AI-generated code samples is used for training and testing the classification model. The system is implemented using Python with Scikit-learn, Tainter, SQLite, Pandas, and NumPy. A graphical user interface enables users to upload source code files or paste code directly into the application. After preprocessing and feature extraction, the trained SVM model predicts the class of the source code and displays the prediction along with a confidence score. The prediction history is stored in an SQLite database for future reference.
📚 How to Cite:
D. Divya, B. Jhansi, S. Guna Priyanka, D.Aruna Padma, V. Sailaja, K. Leela Pavani, Sunil B , DETECTION OF AI GENERATION OF CODE USING MACHINE LEARNING , Volume 12 , Issue 7, July 2026, EPRA International Journal of Multidisciplinary Research (IJMR) , Pages: 911 - 921 ,