The field of code analysis is moving towards more innovative and effective methods for detecting vulnerabilities and clones. Researchers are exploring new approaches that leverage machine learning, graph neural networks, and large language models to improve the accuracy and robustness of vulnerability detection. Notably, the use of Code Property Graphs and explainable attention mechanisms is becoming increasingly popular for providing transparency and trust in security triage. Furthermore, the integration of large language models with code analysis techniques is showing promising results in detecting vulnerabilities and identifying vulnerable code clones. Overall, the field is witnessing a significant shift towards more sophisticated and explainable methods for code analysis. Noteworthy papers include:

• AlphaCC, which proposes a novel framework for code clone detection inspired by AlphaFold.
• ExplainVulD, which presents a graph-based framework for vulnerability detection that achieves high accuracy and produces explainable outputs.
• LLMxCPG, which integrates Code Property Graphs with Large Language Models for robust vulnerability detection and achieves significant improvements in F1-score over state-of-the-art baselines.
• VulCoCo, which proposes a lightweight and scalable approach for detecting vulnerable code clones using embedding-based retrieval and large language model validation.
• RevisitVD, which provides an extensive evaluation of pre-trained language models for vulnerability detection and highlights their strengths and limitations.
• VulPathFinder, which introduces a novel Graph Neural Network model for detecting sink statements and discovering vulnerable paths in open-source code.