Research Papers

Detecting disinformation that blends manipulated text and images has become increasingly challenging, as AI tools make synthetic content easy to generate and disseminate. While most existing AI safety benchmarks focus on single modality misinformation (i.e., false content shared without intent to deceive), intentional multimodal disinformation, such as propaganda or conspiracy theories that imitate credible news, remains largely unaddressed. We introduce the Vision-Language Disinformation Detection Benchmark (VLDBench), the first large-scale resource supporting both unimodal (text-only) and multimodal (text + image) disinformation detection. VLDBench comprises approximately 62,000 labeled text-image pairs across 13 categories, curated from 58 news outlets. Using a semi-automated pipeline followed by expert review, 22 domain experts invested over 500 hours to produce high-quality annotations with substantial inter-annotator agreement. Evaluations of state-of-the-art Large Language Models (LLMs) and Vision-Language Models (VLMs) on VLDBench show that incorporating visual cues improves detection accuracy by 5 to 35 percentage points over text-only models. VLDBench provides data and code for evaluation, fine-tuning, and robustness testing to support disinformation analysis. Developed in alignment with AI governance frameworks (e.g., the MIT AI Risk Repository), VLDBench offers a principled foundation for advancing trustworthy disinformation detection in multimodal media. Project: https://vectorinstitute.github.io/VLDBench/ Dataset: https://huggingface.co/datasets/vector-institute/VLDBench Code: https://github.com/VectorInstitute/VLDBench

The development of artificial general intelligence (AGI) is likely to be one of humanity's most consequential technological advancements. Leading AI labs and scientists have called for the global prioritization of AI safety citing existential risks comparable to nuclear war. However, research on catastrophic risks and AI alignment is often met with skepticism, even by experts. Furthermore, online debate over the existential risk of AI has begun to turn tribal (e.g. name-calling such as "doomer" or "accelerationist"). Until now, no systematic study has explored the patterns of belief and the levels of familiarity with AI safety concepts among experts. I surveyed 111 AI experts on their familiarity with AI safety concepts, key objections to AI safety, and reactions to safety arguments. My findings reveal that AI experts cluster into two viewpoints -- an "AI as controllable tool" and an "AI as uncontrollable agent" perspective -- diverging in beliefs toward the importance of AI safety. While most experts (78%) agreed or strongly agreed that "technical AI researchers should be concerned about catastrophic risks", many were unfamiliar with specific AI safety concepts. For example, only 21% of surveyed experts had heard of "instrumental convergence," a fundamental concept in AI safety predicting that advanced AI systems will tend to pursue common sub-goals (such as self-preservation). The least concerned participants were the least familiar with concepts like this, suggesting that effective communication of AI safety should begin with establishing clear conceptual foundations in the field.

Ensuring Medical AI Safety: Interpretability-Driven Detection and Mitigation of Spurious Model Behavior and Associated Data

Frederik Pahde, Thomas Wiegand, Sebastian Lapuschkin +1

Deep neural networks are increasingly employed in high-stakes medical applications, despite their tendency for shortcut learning in the presence of spurious correlations, which can have potentially fatal consequences in practice. Whereas a multitude of works address either the detection or mitigation of such shortcut behavior in isolation, the Reveal2Revise approach provides a comprehensive bias mitigation framework combining these steps. However, effectively addressing these biases often requires substantial labeling efforts from domain experts. In this work, we review the steps of the Reveal2Revise framework and enhance it with semi-automated interpretability-based bias annotation capabilities. This includes methods for the sample- and feature-level bias annotation, providing valuable information for bias mitigation methods to unlearn the undesired shortcut behavior. We show the applicability of the framework using four medical datasets across two modalities, featuring controlled and real-world spurious correlations caused by data artifacts. We successfully identify and mitigate these biases in VGG16, ResNet50, and contemporary Vision Transformer models, ultimately increasing their robustness and applicability for real-world medical tasks. Our code is available at https://github.com/frederikpahde/medical-ai-safety.

Encrypted network communication ensures confidentiality, integrity, and privacy between endpoints. However, attackers are increasingly exploiting encryption to conceal malicious behavior. Detecting unknown encrypted malicious traffic without decrypting the payloads remains a significant challenge. In this study, we investigate the integration of explainable artificial intelligence (XAI) techniques to detect malicious network traffic. We employ ensemble learning models to identify malicious activity using multi-view features extracted from various aspects of encrypted communication. To effectively represent malicious communication, we compiled a robust dataset with 1,127 unique connections, more than any other available open-source dataset, and spanning 54 malware families. Our models were benchmarked against the CTU-13 dataset, achieving performance of over 99% accuracy, precision, and F1-score. Additionally, the eXtreme Gradient Boosting (XGB) model demonstrated 99.32% accuracy, 99.53% precision, and 99.43% F1-score on our custom dataset. By leveraging Shapley Additive Explanations (SHAP), we identified that the maximum packet size, mean inter-arrival time of packets, and transport layer security version used are the most critical features for the global model explanation. Furthermore, key features were identified as important for local explanations across both datasets for individual traffic samples. These insights provide a deeper understanding of the model decision-making process, enhancing the transparency and reliability of detecting malicious encrypted traffic.

The justice system has increasingly employed AI techniques to enhance efficiency, yet limitations remain in improving the quality of decision-making, particularly regarding transparency and explainability needed to uphold public trust in legal AI. To address these challenges, we propose a large language model based multi-agent framework named AgentsBench, which aims to simultaneously improve both efficiency and quality in judicial decision-making. Our approach leverages multiple LLM-driven agents that simulate the collaborative deliberation and decision making process of a judicial bench. We conducted experiments on legal judgment prediction task, and the results show that our framework outperforms existing LLM based methods in terms of performance and decision quality. By incorporating these elements, our framework reflects real-world judicial processes more closely, enhancing accuracy, fairness, and society consideration. AgentsBench provides a more nuanced and realistic methods of trustworthy AI decision-making, with strong potential for application across various case types and legal scenarios.

In recent years, much research has been dedicated to uncovering the environmental impact of Artificial Intelligence (AI), showing that training and deploying AI systems require large amounts of energy and resources, and the outcomes of AI may lead to decisions and actions that may negatively impact the environment. This new knowledge raises new ethical questions, such as: When is it (un)justifiable to develop an AI system, and how to make design choices, considering its environmental impact? However, so far, the environmental impact of AI has largely escaped ethical scrutiny, as AI ethics tends to focus strongly on themes such as transparency, privacy, safety, responsibility, and bias. Considering the environmental impact of AI from an ethical perspective expands the scope of AI ethics beyond an anthropocentric focus towards including more-than-human actors such as animals and ecosystems. This paper explores the ethical implications of the environmental impact of AI for designing AI systems by drawing on environmental justice literature, in which three categories of justice are distinguished, referring to three elements that can be unjust: the distribution of benefits and burdens (distributive justice), decision-making procedures (procedural justice), and institutionalized social norms (justice as recognition). Based on these tenets of justice, we outline criteria for developing environmentally just AI systems, given their ecological impact.

Machine-generated music (MGM) has become a groundbreaking innovation with wide-ranging applications, such as music therapy, personalised editing, and creative inspiration within the music industry. However, the unregulated proliferation of MGM presents considerable challenges to the entertainment, education, and arts sectors by potentially undermining the value of high-quality human compositions. Consequently, MGM detection (MGMD) is crucial for preserving the integrity of these fields. Despite its significance, MGMD domain lacks comprehensive systematic evaluation results necessary to drive meaningful progress. To address this gap, we conduct experiments on existing large-scale datasets using a range of foundational models for audio processing, establishing systematic evaluation results tailored to the MGMD task. Our selection includes traditional machine learning models, deep neural networks, Transformer-based architectures, and State space models (SSM). Recognising the inherently multimodal nature of music, which integrates both melody and lyrics, we also explore fundamental multimodal models in our experiments. Beyond providing basic binary classification outcomes, we delve deeper into model behaviour using multiple explainable Artificial Intelligence (XAI) tools, offering insights into their decision-making processes. Our analysis reveals that ResNet18 performs the best according to in-domain and out-of-domain tests. By providing a comprehensive comparison of systematic evaluation results and their interpretability, we propose several directions to inspire future research to develop more robust and effective detection methods for MGM. We provide our codes and some samples on Github repository.

Artificial Intelligence (AI) is a broad field that is upturning mental health care in many ways, from addressing anxiety, depression, and stress to increasing access, personalization of treatment, and real-time monitoring that enhances patient outcomes. The current paper discusses the evolution, present application, and future challenges in the field of AI for mental health and well-being. From the early chatbot models, such as ELIZA, to modern machine learning systems, the integration of AI in mental health has grown rapidly to augment traditional treatment and open innovative solutions. AI-driven tools provide continuous support, offering personalized interventions and addressing issues such as treatment access and patient stigma. AI also enables early diagnosis through the analysis of complex datasets, including speech patterns and social media behavior, to detect early signs of conditions like depression and Post-Traumatic Stress Disorder (PTSD). Ethical challenges persist, however, most notably around privacy, data security, and algorithmic bias. With AI at the core of mental health care, there is a dire need to develop strong ethical frameworks that ensure patient rights are protected, access is equitable, and transparency is maintained in AI applications. Going forward, the role of AI in mental health will continue to evolve, and continued research and policy development will be needed to meet the diverse needs of patients while mitigating associated risks.

Natural Language Processing (NLP) research on AI Safety and social bias in AI has focused on safety for humans and social bias against human minorities. However, some AI ethicists have argued that the moral significance of nonhuman animals has been ignored in AI research. Therefore, the purpose of this study is to investigate whether there is speciesism, i.e., discrimination against nonhuman animals, in NLP research. First, we explain why nonhuman animals are relevant in NLP research. Next, we survey the findings of existing research on speciesism in NLP researchers, data, and models and further investigate this problem in this study. The findings of this study suggest that speciesism exists within researchers, data, and models, respectively. Specifically, our survey and experiments show that (a) among NLP researchers, even those who study social bias in AI, do not recognize speciesism or speciesist bias; (b) among NLP data, speciesist bias is inherent in the data annotated in the datasets used to evaluate NLP models; (c) OpenAI GPTs, recent NLP models, exhibit speciesist bias by default. Finally, we discuss how we can reduce speciesism in NLP research.

Formal Explanations for Neuro-Symbolic AI

Sushmita Paul, Jinqiang Yu, Jip J. Dekker +2

Despite the practical success of Artificial Intelligence (AI), current neural AI algorithms face two significant issues. First, the decisions made by neural architectures are often prone to bias and brittleness. Second, when a chain of reasoning is required, neural systems often perform poorly. Neuro-symbolic artificial intelligence is a promising approach that tackles these (and other) weaknesses by combining the power of neural perception and symbolic reasoning. Meanwhile, the success of AI has made it critical to understand its behaviour, leading to the development of explainable artificial intelligence (XAI). While neuro-symbolic AI systems have important advantages over purely neural AI, we still need to explain their actions, which are obscured by the interactions of the neural and symbolic components. To address the issue, this paper proposes a formal approach to explaining the decisions of neuro-symbolic systems. The approach hinges on the use of formal abductive explanations and on solving the neuro-symbolic explainability problem hierarchically. Namely, it first computes a formal explanation for the symbolic component of the system, which serves to identify a subset of the individual parts of neural information that needs to be explained. This is followed by explaining only those individual neural inputs, independently of each other, which facilitates succinctness of hierarchical formal explanations and helps to increase the overall performance of the approach. Experimental results for a few complex reasoning tasks demonstrate practical efficiency of the proposed approach, in comparison to purely neural systems, from the perspective of explanation size, explanation time, training time, model sizes, and the quality of explanations reported.

Artificial intelligence (AI) systems have substantially improved dermatologists' diagnostic accuracy for melanoma, with explainable AI (XAI) systems further enhancing clinicians' confidence and trust in AI-driven decisions. Despite these advancements, there remains a critical need for objective evaluation of how dermatologists engage with both AI and XAI tools. In this study, 76 dermatologists participated in a reader study, diagnosing 16 dermoscopic images of melanomas and nevi using an XAI system that provides detailed, domain-specific explanations. Eye-tracking technology was employed to assess their interactions. Diagnostic performance was compared with that of a standard AI system lacking explanatory features. Our findings reveal that XAI systems improved balanced diagnostic accuracy by 2.8 percentage points relative to standard AI. Moreover, diagnostic disagreements with AI/XAI systems and complex lesions were associated with elevated cognitive load, as evidenced by increased ocular fixations. These insights have significant implications for clinical practice, the design of AI tools for visual tasks, and the broader development of XAI in medical diagnostics.

Causality is vital for understanding true cause-and-effect relationships between variables within predictive models, rather than relying on mere correlations, making it highly relevant in the field of Explainable AI. In an automated decision-making scenario, causal inference methods can analyze the underlying data-generation process, enabling explanations of a model's decision by manipulating features and creating counterfactual examples. These counterfactuals explore hypothetical scenarios where a minimal number of factors are altered, providing end-users with valuable information on how to change their situation. However, interpreting a set of multiple counterfactuals can be challenging for end-users who are not used to analyzing raw data records. In our work, we propose a novel multi-step pipeline that uses counterfactuals to generate natural language explanations of actions that will lead to a change in outcome in classifiers of tabular data using LLMs. This pipeline is designed to guide the LLM through smaller tasks that mimic human reasoning when explaining a decision based on counterfactual cases. We conducted various experiments using a public dataset and proposed a method of closed-loop evaluation to assess the coherence of the final explanation with the counterfactuals, as well as the quality of the content. Results are promising, although further experiments with other datasets and human evaluations should be carried out.