Although deep neural networks (DNNs) have achieved a great success in various computer vision tasks, it is recently found that they are vulnerable to adversarial attacks. In this paper, we focus on the so-called \textit{backdoor attack}, which injects a backdoor trigger to a small portion of training data (also known as data poisoning) such that the trained DNN induces misclassification while facing examples with this trigger. To be specific, we carefully study the effect of both real and synthetic backdoor attacks on the internal response of vanilla and backdoored DNNs through the lens of Gard-CAM. Moreover, we show that the backdoor attack induces a significant bias in neuron activation in terms of the $\ell_\infty$ norm of an activation map compared to its $\ell_1$ and $\ell_2$ norm. Spurred by our results, we propose the \textit{$\ell_\infty$-based neuron pruning} to remove the backdoor from the backdoored DNN. Experiments show that our method could effectively decrease the attack success rate, and also hold a high classification accuracy for clean images.

Fine-tuning Large Language Models (LLMs) on some task-specific datasets has been a primary use of LLMs. However, it has been empirically observed that this approach to enhancing capability inevitably compromises safety, a phenomenon also known as the safety-capability trade-off in LLM fine-tuning. This paper presents a theoretical framework for understanding the interplay between safety and capability in two primary safety-aware LLM fine-tuning strategies, providing new insights into the effects of data similarity, context overlap, and alignment loss landscape. Our theoretical results characterize the fundamental limits of the safety-capability trade-off in LLM fine-tuning, which are also validated by numerical experiments.

Deepfakes have become a universal and rapidly intensifying concern of generative AI across various media types such as images, audio, and videos. Among these, audio deepfakes have been of particular concern due to the ease of high-quality voice synthesis and distribution via platforms such as social media and robocalls. Consequently, detecting audio deepfakes plays a critical role in combating the growing misuse of AI-synthesized speech. However, real-world scenarios often introduce various audio corruptions, such as noise, modification, and compression, that may significantly impact detection performance. This work systematically evaluates the robustness of 10 audio deepfake detection models against 16 common corruptions, categorized into noise perturbation, audio modification, and compression. Using both traditional deep learning models and state-of-the-art foundation models, we make four unique observations. First, our findings show that while most models demonstrate strong robustness to noise, they are notably more vulnerable to modifications and compression, especially when neural codecs are applied. Second, speech foundation models generally outperform traditional models across most scenarios, likely due to their self-supervised learning paradigm and large-scale pre-training. Third, our results show that increasing model size improves robustness, albeit with diminishing returns. Fourth, we demonstrate how targeted data augmentation during training can enhance model resilience to unseen perturbations. A case study on political speech deepfakes highlights the effectiveness of foundation models in achieving high accuracy under real-world conditions. These findings emphasize the importance of developing more robust detection frameworks to ensure reliability in practical deployment settings.

Differentially private (DP) synthetic data has become the de facto standard for releasing sensitive data. However, many DP generative models suffer from the low utility of synthetic data, especially for high-resolution images. On the other hand, one of the emerging techniques in parameter efficient fine-tuning (PEFT) is visual prompting (VP), which allows well-trained existing models to be reused for the purpose of adapting to subsequent downstream tasks. In this work, we explore such a phenomenon in constructing captivating generative models with DP constraints. We show that VP in conjunction with DP-NTK, a DP generator that exploits the power of the neural tangent kernel (NTK) in training DP generative models, achieves a significant performance boost, particularly for high-resolution image datasets, with accuracy improving from 0.644$\pm$0.044 to 0.769. Lastly, we perform ablation studies on the effect of different parameters that influence the overall performance of VP-NTK. Our work demonstrates a promising step forward in improving the utility of DP synthetic data, particularly for high-resolution images.

As the open-weight AI landscape continues to proliferate-with model development, significant investment, and user interest-it becomes increasingly important to predict which models will ultimately drive innovation and shape AI ecosystems. Building on parallels with citation dynamics in scientific literature, we propose a framework to quantify how an open-weight model's influence evolves. Specifically, we adapt the model introduced by Wang et al. for scientific citations, using three key parameters-immediacy, longevity, and relative fitness-to track the cumulative number of fine-tuned models of an open-weight model. Our findings reveal that this citation-style approach can effectively capture the diverse trajectories of open-weight model adoption, with most models fitting well and outliers indicating unique patterns or abrupt jumps in usage.

Backdoor attacks on deep neural networks (DNNs) have emerged as a significant security threat, allowing adversaries to implant hidden malicious behaviors during the model training phase. Pre-processing-based defense, which is one of the most important defense paradigms, typically focuses on input transformations or backdoor trigger inversion (BTI) to deactivate or eliminate embedded backdoor triggers during the inference process. However, these methods suffer from inherent limitations: transformation-based defenses often fail to balance model utility and defense performance, while BTI-based defenses struggle to accurately reconstruct trigger patterns without prior knowledge. In this paper, we propose REFINE, an inversion-free backdoor defense method based on model reprogramming. REFINE consists of two key components: \textbf{(1)} an input transformation module that disrupts both benign and backdoor patterns, generating new benign features; and \textbf{(2)} an output remapping module that redefines the model's output domain to guide the input transformations effectively. By further integrating supervised contrastive loss, REFINE enhances the defense capabilities while maintaining model utility. Extensive experiments on various benchmark datasets demonstrate the effectiveness of our REFINE and its resistance to potential adaptive attacks.

As large language models (LLMs) become integrated into everyday applications, ensuring their robustness and security is increasingly critical. In particular, LLMs can be manipulated into unsafe behaviour by prompts known as jailbreaks. The variety of jailbreak styles is growing, necessitating the use of external defences known as guardrails. While many jailbreak defences have been proposed, not all defences are able to handle new out-of-distribution attacks due to the narrow segment of jailbreaks used to align them. Moreover, the lack of systematisation around defences has created significant gaps in their practical application. In this work, we perform systematic benchmarking across 15 different defences, considering a broad swathe of malicious and benign datasets. We find that there is significant performance variation depending on the style of jailbreak a defence is subject to. Additionally, we show that based on current datasets available for evaluation, simple baselines can display competitive out-of-distribution performance compared to many state-of-the-art defences. Code is available at https://github.com/IBM/Adversarial-Prompt-Evaluation.

AI safety is a rapidly growing area of research that seeks to prevent the harm and misuse of frontier AI technology, particularly with respect to generative AI (GenAI) tools that are capable of creating realistic and high-quality content through text prompts. Examples of such tools include large language models (LLMs) and text-to-image (T2I) diffusion models. As the performance of various leading GenAI models approaches saturation due to similar training data sources and neural network architecture designs, the development of reliable safety guardrails has become a key differentiator for responsibility and sustainability. This paper presents a formalization of the concept of computational safety, which is a mathematical framework that enables the quantitative assessment, formulation, and study of safety challenges in GenAI through the lens of signal processing theory and methods. In particular, we explore two exemplary categories of computational safety challenges in GenAI that can be formulated as hypothesis testing problems. For the safety of model input, we show how sensitivity analysis and loss landscape analysis can be used to detect malicious prompts with jailbreak attempts. For the safety of model output, we elucidate how statistical signal processing and adversarial learning can be used to detect AI-generated content. Finally, we discuss key open research challenges, opportunities, and the essential role of signal processing in computational AI safety. Signal processing has played a pivotal role in ensuring the stability, security, and efficiency of numerous engineering systems and information technologies, including, but not limited to, telecommunications, information forensics and security, machine learning, data science, and control systems. With the recent advances, wide accessibility, and deep integration of generative AI (GenAI) tools into our society and technology, such as ChatGPT and the emerging agentic AI applications, understanding and mitigating the associated risks of the so-called "frontier AI technology" is essential to ensure a responsible and sustainable use of GenAI. In addition, as the performance of state-ofthe-art GanAI models surpasses that of an average human in certain tasks, but reaches a plateau in standardized capability evaluation benchmarks due to similar training data sources and neural network architecture design (e.g., the use of decoder-only transformers), improving and ensuring safety is becoming the new arms race among GenAI stakeholders. EU AI Act, AI safety institutes, etc.), there are growing concerns about the broader socio-technical impacts [1].

Model merging is an efficient way of obtaining a multi-task model from several pretrained models without further fine-tuning, and it has gained attention in various domains, including natural language processing (NLP). Despite the efficiency, a key challenge in model merging is the seemingly inevitable decrease in task performance as the number of models increases. In this paper, we propose $\mathbf{S}$pectral $\mathbf{T}$runcation $\mathbf{A}$nd $\mathbf{R}$escale (STAR) that aims at mitigating ``merging conflicts'' by truncating small components in the respective spectral spaces, which is followed by an automatic parameter rescaling scheme to retain the nuclear norm of the original matrix. STAR requires no additional inference on original training data and is robust to hyperparamater choice. We demonstrate the effectiveness of STAR through extensive model merging cases on diverse NLP tasks. Specifically, STAR works robustly across varying model sizes, and can outperform baselines by 4.2$\%$ when merging 12 models on Flan-T5. Our code is publicly available at https://github.com/IBM/STAR.

The advancement of large language models (LLMs) has made it difficult to differentiate human-written text from AI-generated text. Several AI-text detectors have been developed in response, which typically utilize a fixed global threshold (e.g., {\theta} = 0.5) to classify machine-generated text. However, we find that one universal threshold can fail to account for subgroup-specific distributional variations. For example, when using a fixed threshold, detectors make more false positive errors on shorter human-written text than longer, and more positive classifications on neurotic writing styles than open among long text. These discrepancies can lead to misclassification that disproportionately affects certain groups. We address this critical limitation by introducing FairOPT, an algorithm for group-specific threshold optimization in AI-generated content classifiers. Our approach partitions data into subgroups based on attributes (e.g., text length and writing style) and learns decision thresholds for each group, which enables careful balancing of performance and fairness metrics within each subgroup. In experiments with four AI text classifiers on three datasets, FairOPT enhances overall F1 score and decreases balanced error rate (BER) discrepancy across subgroups. Our framework paves the way for more robust and fair classification criteria in AI-generated output detection.