Phishing attacks are a major threat to online security, exploiting user vulnerabilities to steal sensitive information. Various methods have been developed to counteract phishing, each with varying levels of accuracy, but they also face notable limitations. In this study, we introduce PhishAgent, a multimodal agent that combines a wide range of tools, integrating both online and offline knowledge bases with Multimodal Large Language Models (MLLMs). This combination leads to broader brand coverage, which enhances brand recognition and recall. Furthermore, we propose a multimodal information retrieval framework designed to extract the relevant top k items from offline knowledge bases, using available information from a webpage, including logos and HTML. Our empirical results, based on three real-world datasets, demonstrate that the proposed framework significantly enhances detection accuracy and reduces both false positives and false negatives, while maintaining model efficiency. Additionally, PhishAgent shows strong resilience against various types of adversarial attacks.

Early screening for prenatal depression is of critical importance to maternal and infant health. However, traditional screening methods are fundamentally limited by a confluence of challenges, including data imbalance, resource constraints, poor model adaptability, and a lack of interpretability. While Large Language Models (LLMs) show potential in medical text analysis, their direct application to class-imbalanced tasks requires significant optimization. To address these limitations, we propose and evaluate an Instruction Fine-Tuning methodology, integrating prompt engineering directly into the training phase by prepending each questionnaire-based case description with a structured, task-specific directive. This directive encodes expert domain knowledge and an explicit instruction for the model to prioritize high recall, thereby directly addressing the clinical need to minimize false negatives. Several language models were adapted using this instruction-augmented dataset via the parameter-efficient Low-Rank Adaptation (LoRA) technique. The experimental results demonstrate that models trained with our Instruction Fine-Tuning method exhibit significantly higher recall for the minority class and achieve a more balanced overall performance. Our best-performing instruction-tuned model achieved an AUC of 0.81 and a Balanced Accuracy of 0.72, validating the effectiveness of our approach. Crucially, our interpretability analysis revealed that the models learned to focus on clinically-relevant risk factors, providing transparent and plausible justifications for their predictions. This research confirms that Instruction Fine-Tuning is a highly effective methodology for adapting LLMs to sensitive, imbalanced clinical screening tasks. The proposed approach presents a robust, interpretable, and adaptable solution well-suited to the demands of low-resource medical environments.

Large language models (LLMs) have shown remarkable performance across a range of NLP tasks.However, their strong instructionfollowing capabilities and inability to distinguish instructions from data content make them vulnerable to indirect prompt injection attacks.In such attacks, instructions with malicious purposes are injected into external data sources, such as web documents.When LLMs retrieve this injected data through tools, such as a search engine and execute the injected instructions, they provide misled responses.Recent attack methods have demonstrated potential, but their abrupt instruction injection often undermines their effectiveness.Motivated by the limitations of existing attack methods, we propose Topi-cAttack, which prompts the LLM to generate a fabricated conversational transition prompt that gradually shifts the topic toward the injected instruction, making the injection smoother and enhancing the plausibility and success of the attack.Through comprehensive experiments, TopicAttack achieves state-of-the-art performance, with an attack success rate (ASR) over 90% in most cases, even when various defense methods are applied.We further analyze its effectiveness by examining attention scores.We find that a higher injected-to-original attention ratio leads to a greater success probability, and our method achieves a much higher ratio than the baseline methods. 1

COVID-19 has been rapidly spreading and causing hundreds of thousands of casualties across the world. This disease originated in Wuhan, China in 2019, and it has since proliferated into a pandemic, causing the public's daily lives to change drastically around the world. Despite the evolution of COVID-19, remarkably little is known about the efficacy of proposed protective safety orders. Thus, our paper is intended to study the effectiveness of the local and state government restrictions and closures in limiting the spread of COVID-19. To mathematically model the spread of COVID-19, we propose a time-dependent SIR model together with Lasso to monitor the trajectories of the transmission and recover rates in relation to the government closures and restrictions, and further predict the number of cases. To validate our model, we conduct both simulation and state level data analysis. Since the government orders vary among different states, we use Texas as an example state to illustrate our algorithm for assessing the government intervention. Our one-day prediction error for the confirmed cases is around 2.47% and less than 1% for the recovered cases. We also find that there are many intervention methods that corresponded to changes in infection rate and recovery rate of the population.