The field of software bug detection and repair is rapidly advancing with the application of machine learning and graph neural networks. Recent research has focused on improving the accuracy of bug detection models by incorporating semantic relationships between changed code lines and leveraging large language models to generate annotations for memory leak detection. Additionally, novel approaches to fuzz testing have been proposed, including the use of relational graph convolutional networks to detect root-cause deletion lines and the utilization of large language models to generate targeted test harnesses. These developments have the potential to significantly improve the efficiency and effectiveness of software development and testing. Noteworthy papers in this area include: Identifying Root Cause of bugs by Capturing Changed Code Lines with Relational Graph Neural Networks, which improved recall and MFR metrics by up to 24.536%. Detecting the Root Cause Code Lines in Bug-Fixing Commits by Heterogeneous Graph Learning, which achieved significant improvements of up to 96.83% in MFR compared to state-of-the-art approaches. Directed Greybox Fuzzing via Large Language Model, which successfully triggered 17 real-world vulnerabilities and discovered 9 previously unknown vulnerabilities.