Cyber Threat Intelligence (CTI) summarization task requires the system to generate concise and accurate highlights from raw intelligence data, which plays an important role in providing decision-makers with crucial information to quickly detect and respond to cyber threats in the cybersecurity domain. However, efficient techniques for summarizing CTI reports, including facts, analytical insights, attack processes, etc., have largely been unexplored, primarily due to the lack of available dataset. To this end, we present CTISum, a new benchmark for CTI summarization task. Considering the importance of attack process, a novel fine-grained subtask of attack process summarization is proposed to enable defenders to assess risk, identify security gaps, vulnerabilities, and so on. Specifically, we first design a multi-stage annotation pipeline to gather and annotate the CTI data, and then benchmark the CTISum with a collection of extractive and abstractive summarization methods. Experimental results show that current state-of-the-art models exhibit limitations when applied to CTISum, underscoring the fact that automatically producing concise summaries of CTI reports remains an open research challenge.

Explaining decisions of black-box classifiers is both important and computationally challenging. In this paper, we scrutinize explainers that generate feature-based explanations from samples or datasets. We start by presenting a set of desirable properties that explainers would ideally satisfy, delve into their relationships, and highlight incompatibilities of some of them. We identify the entire family of explainers that satisfy two key properties which are compatible with all the others. Its instances provide sufficient reasons, called weak abductive explanations.We then unravel its various subfamilies that satisfy subsets of compatible properties. Indeed, we fully characterize all the explainers that satisfy any subset of compatible properties. In particular, we introduce the first (broad family of) explainers that guarantee the existence of explanations and their global consistency.We discuss some of its instances including the irrefutable explainer and the surrogate explainer whose explanations can be found in polynomial time.

Large language models (LLMs) have demonstrated remarkable capabilities but often struggle to align with human preferences, leading to harmful or undesirable outputs. Preference learning, which trains models to distinguish between preferred and non-preferred responses based on human feedback, has become a crucial component for ensuring that LLMs align with human values. Despite the widespread adoption in real-world systems, a thorough theoretical understanding of the generalization guarantees for these models remain lacking. This paper bridges that gap by introducing a new theoretical framework to analyze the generalization guarantees of models trained with direct preference optimization (DPO). While existing generalization theory often focuses on overparameterized models achieving near-optimal loss or models independent of the training process, our framework rigorously assesses how well models generalize after a finite number of gradient steps, reflecting real-world LLM training practices. By analyzing the reward margin associated with each sample and its trajectory throughout training, we can effectively bound the generalization error. We derive learning guarantees showing that, under specific conditions, models trained with DPO can correctly discern preferred responses on unseen data with high probability. These insights are empirically validated on contemporary LLMs, underscoring the practical relevance of our theoretical findings.

In recent years, the convergence of cybersecurity, artificial intelligence (AI), and data management has emerged as a critical area of research, driven by the increasing complexity and interdependence of modern technological ecosystems. This paper provides a comprehensive review and analysis of integrative approaches that harness AI techniques to enhance cybersecurity frameworks and optimize data management practices. By exploring the synergies between these domains, we identify key trends, challenges, and future directions that hold the potential to revolutionize the way organizations protect, analyze, and leverage their data. Our findings highlight the necessity of cross-disciplinary strategies that incorporate AI-driven automation, real-time threat detection, and advanced data analytics to build more resilient and adaptive security architectures.

Product classification is a crucial task in international trade, as compliance regulations are verified and taxes and duties are applied based on product categories. Manual classification of products is time-consuming and error-prone, and the sheer volume of products imported and exported renders the manual process infeasible. Consequently, e-commerce platforms and enterprises involved in international trade have turned to automatic product classification using machine learning. However, current approaches do not consider the real-world challenges associated with product classification, such as very abbreviated and incomplete product descriptions. In addition, recent advancements in generative Large Language Models (LLMs) and their reasoning capabilities are mainly untapped in product classification and e-commerce. In this research, we explore the real-life challenges of industrial classification and we propose data perturbations that allow for realistic data simulation. Furthermore, we employ LLM-based product classification to improve the robustness of the prediction in presence of incomplete data. Our research shows that LLMs with in-context learning outperform the supervised approaches in the clean-data scenario. Additionally, we illustrate that LLMs are significantly more robust than the supervised approaches when data attacks are present.

Eliciting harmful behavior from large language models (LLMs) is an important task to ensure the proper alignment and safety of the models. Often when training LLMs, ethical guidelines are followed yet alignment failures may still be uncovered through red teaming adversarial attacks. This work frames the red-teaming problem as a Markov decision process (MDP) and uses Monte Carlo tree search to find harmful behaviors of black-box, closed-source LLMs. We optimize token-level prompt suffixes towards targeted harmful behaviors on white-box LLMs and include a naturalistic loss term, log-perplexity, to generate more natural language attacks for better interpretability. The proposed algorithm, Kov, trains on white-box LLMs to optimize the adversarial attacks and periodically evaluates responses from the black-box LLM to guide the search towards more harmful black-box behaviors. In our preliminary study, results indicate that we can jailbreak black-box models, such as GPT-3.5, in only 10 queries, yet fail on GPT-4$-$which may indicate that newer models are more robust to token-level attacks. All work to reproduce these results is open sourced (https://github.com/sisl/Kov.jl).

Interpreting and assessing goal driven actions is vital to understanding and reasoning over complex events. It is important to be able to acquire the knowledge needed for this understanding, though doing so is challenging. We argue that such knowledge can be elicited through a participant achievement lens. We analyze a complex event in a narrative according to the intended achievements of the participants in that narrative, the likely future actions of the participants, and the likelihood of goal success. We collect 6.3K high quality goal and action annotations reflecting our proposed participant achievement lens, with an average weighted Fleiss-Kappa IAA of 80%. Our collection contains annotated alternate versions of each narrative. These alternate versions vary minimally from the "original" story, but can license drastically different inferences. Our findings suggest that while modern large language models can reflect some of the goal-based knowledge we study, they find it challenging to fully capture the design and intent behind concerted actions, even when the model pretraining included the data from which we extracted the goal knowledge. We show that smaller models fine-tuned on our dataset can achieve performance surpassing larger models.

A natural question is how LLMs can be applied to the cybersecurity domain, specifically for generating As the complexity of modern systems increases, so does the importance attack graphs. of assessing their security posture through effective vulnerability The aim of this paper is to investigate the potential of using large management and threat modeling techniques. One powerful language models such as ChatGPT for automating the generation tool in the arsenal of cybersecurity professionals is the attack graph, of attack graphs. Our approach leverages LLM capabilities to understand a representation of all potential attack paths within a system that an and chain Common Vulnerabilities and Exposures (CVEs) adversary might exploit to achieve a certain objective. Traditional based on their preconditions and postconditions. By interpreting methods of generating attack graphs involve expert knowledge, CVE descriptions and associated metadata, LLMs can generate links manual curation, and computational algorithms that might not between vulnerabilities, offering a dynamic way to visualize possible cover the entire threat landscape due to the ever-evolving nature attack vectors. In addition, this paper explores using LLMs for of vulnerabilities and exploits. This paper explores the approach generating attack graphs based on textual threat reports, which are of leveraging large language models (LLMs), such as ChatGPT, to often rich sources of data but require manual analysis to transform automate the generation of attack graphs by intelligently chaining into actionable insights. Common Vulnerabilities and Exposures (CVEs) based on their preconditions Our work makes several contributions: and effects. It also shows how to utilize LLMs to create attack graphs from threat reports.

Time roots in applying language models for biomedical applications: models are trained on historical data and will be deployed for new or future data, which may vary from training data. While increasing biomedical tasks have employed state-of-the-art language models, there are very few studies have examined temporal effects on biomedical models when data usually shifts across development and deployment. This study fills the gap by statistically probing relations between language model performance and data shifts across three biomedical tasks. We deploy diverse metrics to evaluate model performance, distance methods to measure data drifts, and statistical methods to quantify temporal effects on biomedical language models. Our study shows that time matters for deploying biomedical language models, while the degree of performance degradation varies by biomedical tasks and statistical quantification approaches. We believe this study can establish a solid benchmark to evaluate and assess temporal effects on deploying biomedical language models.

This paper extends recent investigations on the emotional reasoning abilities of Large Language Models (LLMs). Current research on LLMs has not directly evaluated the distinction between how LLMs predict the self-attribution of emotions and the perception of others' emotions. We first look at carefully crafted emotion-evoking stimuli, originally designed to find patterns of brain neural activity representing fine-grained inferred emotional attributions of others. We show that GPT-4 is especially accurate in reasoning about such stimuli. This suggests LLMs agree with humans' attributions of others' emotions in stereotypical scenarios remarkably more than self-attributions of emotions in idiosyncratic situations. To further explore this, our second study utilizes a dataset containing annotations from both the author and a third-person perspective. We find that GPT-4's interpretations align more closely with human judgments about the emotions of others than with self-assessments. Notably, conventional computational models of emotion primarily rely on self-reported ground truth as the gold standard. However, an average observer's standpoint, which LLMs appear to have adopted, might be more relevant for many downstream applications, at least in the absence of individual information and adequate safety considerations.