The recent advancements in the field of explainable artificial intelligence (XAI) have significantly enhanced the interpretability and reliability of machine learning models, particularly in complex domains such as computer vision and natural language processing. A notable trend is the development of novel evaluation frameworks for attribution maps in convolutional neural networks (CNNs), which aim to provide more robust and consistent interpretations of model predictions. These frameworks often leverage adversarial perturbations and advanced metrics to correct distribution shifts and improve the faithfulness of attribution maps. Additionally, there is a growing focus on improving the interpretability of transformer models, with new methods addressing gradient flow imbalances and enhancing the completeness and faithfulness of explanations. In the realm of mobile app security and testing, XAI techniques are being applied to machine learning models for malware detection, offering insights into decision-making processes and enhancing trust. Furthermore, large language models (LLMs) are being utilized innovatively for tasks such as crowdsourced test report prioritization and automated test transfer across mobile apps, significantly reducing manual effort and improving efficiency. The integration of AI in healthcare diagnostics is also advancing, with smartphone-based solutions for interpreting rapid diagnostic test kits, enhancing accessibility and accuracy. Overall, the field is moving towards more transparent, reliable, and user-friendly AI systems, with a strong emphasis on the faithfulness and interpretability of model predictions.