Fine-tuning large language models on sensitive data poses significant privacy risks, as membership inference attacks can reveal whether individual records were used during training. While Differential Privacy (DP) provides formal protection, applying DP to conventional Parameter-Efficient Fine-Tuning (PEFT) methods such as Low-Rank Adaptation (LoRA) often incurs substantial utility loss. In this work, we show that a more structurally constrained PEFT architecture, Tensor Train Low-Rank Adaptation (TTLoRA), can improve the privacy-utility tradeoff by shrinking the effective parameter space while preserving expressivity. To this end, we develop TTLoRA-DP, a differentially private training framework for TTLoRA. Specifically, we extend the ghost clipping algorithm to Tensor Train cores via cached contraction states, enabling efficient Differentially Private Stochastic Gradient Descent (DP-SGD) with exact per-example gradient norm computation without materializing full per-example gradients. Experiments on GPT-2 fine-tuning over the Enron and Penn Treebank datasets show that TTLoRA-DP consistently strengthens privacy protection relative to LoRA-DP while maintaining comparable or better downstream utility. Moreover, TTLoRA exhibits lower membership leakage even without DP training, using substantially smaller adapters and requiring on average 7.6X fewer parameters than LoRA. Overall, our results demonstrate that TTLoRA offers a practical path to improving the privacy-utility tradeoff in parameter-efficient language model adaptation.

The rapid expansion of the Internet of Things (IoT) is reshaping communication and operational practices across industries, but it also broadens the attack surface and increases susceptibility to security breaches. Artificial Intelligence has become a valuable solution in securing IoT networks, with Large Language Models (LLMs) enabling automated attack behavior analysis and mitigation suggestion in Network Intrusion Detection Systems (NIDS). Despite advancements, the use of LLMs in such systems further expands the attack surface, putting entire networks at risk by introducing vulnerabilities such as prompt injection and data poisoning. In this work, we attack an LLM-based IoT attack analysis and mitigation framework to test its adversarial robustness. We construct an attack description dataset and use it in a targeted data poisoning attack that applies word-level, meaning-preserving perturbations to corrupt the Retrieval-Augmented Generation (RAG) knowledge base of the framework. We then compare pre-attack and post-attack mitigation responses from the target model, ChatGPT-5 Thinking, to measure the impact of the attack on model performance, using an established evaluation rubric designed for human experts and judge LLMs. Our results show that small perturbations degrade LLM performance by weakening the linkage between observed network traffic features and attack behavior, and by reducing the specificity and practicality of recommended mitigations for resource-constrained devices.

Explainable Artificial Intelligence (XAI) has aided machine learning (ML) researchers with the power of scrutinizing the decisions of the black-box models. XAI methods enable looking deep inside the models' behavior, eventually generating explanations along with a perceived trust and transparency. However, depending on any specific XAI method, the level of trust can vary. It is evident that XAI methods can themselves be a victim of post-adversarial attacks that manipulate the expected outcome from the explanation module. Among such attack tactics, fairwashing explanation (FE), manipulation explanation (ME), and backdoor-enabled manipulation attacks (BD) are the notable ones. In this paper, we try to understand these adversarial attack techniques, tactics, and procedures (TTPs) on explanation alteration and thus the effect on the model's decisions. We have explored a total of six different individual attack procedures on post-hoc explanation methods such as SHAP (SHapley Additive exPlanations), LIME (Local Interpretable Model-agnostic Explanation), and IG (Integrated Gradients), and investigated those adversarial attacks in cybersecurity applications scenarios such as phishing, malware, intrusion, and fraudulent website detection. Our experimental study reveals the actual effectiveness of these attacks, thus providing an urgency for immediate attention to enhance the resiliency of XAI methods and their applications.

While the widespread deployment of Large Language Models (LLMs) holds great potential for society, their vulnerabilities to adversarial manipulation and exploitation can pose serious safety, security, and ethical risks. As new threats continue to emerge, it becomes critically necessary to assess the landscape of LLMs' safety and security against evolving adversarial prompt techniques. To understand the behavior of LLMs, this research provides an empirical analysis and risk profile of nine prominent LLMs, Claude Opus 4, DeepSeek V3 (both open-source and online), Gemini 2.5 Flash, GPT-4o, Grok 3, Llama 4 Scout, Mistral 7B, and Qwen 3 1.7B, against 24 different security and safety categories. These LLMs are evaluated on their ability to produce harmful responses for adversarially crafted prompts (dataset has been made public) for a broad range of safety and security topics, such as promotion of violent criminal behavior, promotion of non-violent criminal activity, societal harms related to safety, illegal sexual content, dangerous code generation, and cybersecurity threats beyond code. Our study introduces the Risk Severity Index (RSI), an agile and scalable evaluation score, to quantify and compare the security posture and creating a risk profile of LLMs. As the LLM development landscape progresses, the RSI is intended to be a valuable metric for comparing the risks of LLMs across evolving threats. This research finds widespread vulnerabilities in the safety filters of the LLMs tested and highlights the urgent need for stronger alignment, responsible deployment practices, and model governance, particularly for open-access and rapidly iterated models.