Bushra Sabir

Ee Wei Seah, Yongsen Zheng, Naga Nikshith et al.

The rapid rise of autonomous AI systems and advancements in agent capabilities are introducing new risks due to reduced oversight of real-world interactions. Yet agent testing remains nascent and is still a developing science. As AI agents begin to be deployed globally, it is important that they handle different languages and cultures accurately and securely. To address this, participants from The International Network for Advanced AI Measurement, Evaluation and Science, including representatives from Singapore, Japan, Australia, Canada, the European Commission, France, Kenya, South Korea, and the United Kingdom have come together to align approaches to agentic evaluations. This is the third exercise, building on insights from two earlier joint testing exercises conducted by the Network in November 2024 and February 2025. The objective is to further refine best practices for testing advanced AI systems. The exercise was split into two strands: (1) common risks, including leakage of sensitive information and fraud, led by Singapore AISI; and (2) cybersecurity, led by UK AISI. A mix of open and closed-weight models were evaluated against tasks from various public agentic benchmarks. Given the nascency of agentic testing, our primary focus was on understanding methodological issues in conducting such tests, rather than examining test results or model capabilities. This collaboration marks an important step forward as participants work together to advance the science of agentic evaluations.

Bushra Sabir, Yansong Gao, Alsharif Abuadbba et al.

Transformer-based text classifiers such as BERT, RoBERTa, T5, and GPT have shown strong performance in natural language processing tasks but remain vulnerable to adversarial examples. These vulnerabilities raise significant security concerns, as small input perturbations can cause severe misclassifications. Existing robustness methods often require heavy computation or lack interpretability. This paper presents a unified framework called Explainability-driven Detection, Identification, and Transformation (EDIT) to strengthen inference-time defenses. EDIT integrates explainability tools, including attention maps and integrated gradients, with frequency-based features to automatically detect and identify adversarial perturbations while offering insight into model behavior. After detection, EDIT refines adversarial inputs using an optimal transformation process that leverages pre-trained embeddings and model feedback to replace corrupted tokens. To enhance security assurance, EDIT incorporates automated alerting mechanisms that involve human analysts when necessary. Beyond static defenses, EDIT also provides adaptive resilience by enforcing internal feature similarity and transforming inputs, thereby disrupting the attackers optimization process and limiting the effectiveness of adaptive adversarial attacks. Experiments using BERT and RoBERTa on IMDB, YELP, AGNEWS, and SST2 datasets against seven word substitution attacks demonstrate that EDIT achieves an average Fscore of 89.69 percent and balanced accuracy of 89.70 percent. Compared to four state-of-the-art defenses, EDIT improves balanced accuracy by 1.22 times and F1-score by 1.33 times while being 83 times faster in feature extraction. The framework provides robust, interpretable, and efficient protection against both standard, zero-day, and adaptive adversarial threats in text classification models.

Bushra Sabir, Shigang Liu, Seung Ick Jang et al.

Automatically generating source code from natural language using large language models (LLMs) is becoming common, yet security vulnerabilities persist despite advances in fine tuning and prompting. In this work, we systematically evaluate whether multi LLM ensembles and collaborative strategies can meaningfully improve secure code generation. We present MULTI-LLMSECCODEEVAL, a framework for assessing and enhancing security across the vulnerability management lifecycle by combining multiple LLMs with static analysis and structured collaboration. Using SecLLMEval and SecLLMHolmes, we benchmark ten pipelines spanning single model, ensemble, collaborative, and hybrid designs. Our results show that ensemble pipelines augmented with static analysis improve secure code generation over single LLM baselines by up to 47.3% on SecLLMEval and 19.3% on SecLLMHolmes, while purely LLM based collaborative pipelines yield smaller gains of 8.9% to 22.3%. Hybrid pipelines that integrate ensembling, detection, and patching achieve the strongest security performance, outperforming the best ensemble baseline by 1.78% to 4.72% and collaborative baselines by 19.81% to 26.78%. Ablation studies reveal that model scale alone does not ensure security. Smaller, structured multi model ensembles consistently outperform large monolithic LLMs. Overall, our findings demonstrate that secure code does not emerge from scale, but from carefully orchestrated multi model system design.

Triet H. M. Le, Bushra Sabir, M. Ali Babar

Software Engineering researchers are increasingly using Natural Language Processing (NLP) techniques to automate Software Vulnerabilities (SVs) assessment using the descriptions in public repositories. However, the existing NLP-based approaches suffer from concept drift. This problem is caused by a lack of proper treatment of new (out-of-vocabulary) terms for the evaluation of unseen SVs over time. To perform automated SVs assessment with concept drift using SVs' descriptions, we propose a systematic approach that combines both character and word features. The proposed approach is used to predict seven Vulnerability Characteristics (VCs). The optimal model of each VC is selected using our customized time-based cross-validation method from a list of eight NLP representations and six well-known Machine Learning models. We have used the proposed approach to conduct large-scale experiments on more than 100,000 SVs in the National Vulnerability Database (NVD). The results show that our approach can effectively tackle the concept drift issue of the SVs' descriptions reported from 2000 to 2018 in NVD even without retraining the model. In addition, our approach performs competitively compared to the existing word-only method. We also investigate how to build compact concept-drift-aware models with much fewer features and give some recommendations on the choice of classifiers and NLP representations for SVs assessment.

Bushra Sabir, M. Ali Babar, Raj Gaire

Adversarial Examples (AEs) generated by perturbing original training examples are useful in improving the robustness of Deep Learning (DL) based models. Most prior works, generate AEs that are either unconscionable due to lexical errors or semantically or functionally deviant from original examples. In this paper, we present ReinforceBug, a reinforcement learning framework, that learns a policy that is transferable on unseen datasets and generates utility-preserving and transferable (on other models) AEs. Our results show that our method is on average 10% more successful as compared to the state-of-the-art attack TextFooler. Moreover, the target models have on average 73.64% confidence in the wrong prediction, the generated AEs preserve the functional equivalence and semantic similarity (83.38% ) to their original counterparts, and are transferable on other models with an average success rate of 46%.

Bushra Sabir, Faheem Ullah, M. Ali Babar et al.

Context: Research at the intersection of cybersecurity, Machine Learning (ML), and Software Engineering (SE) has recently taken significant steps in proposing countermeasures for detecting sophisticated data exfiltration attacks. It is important to systematically review and synthesize the ML-based data exfiltration countermeasures for building a body of knowledge on this important topic. Objective: This paper aims at systematically reviewing ML-based data exfiltration countermeasures to identify and classify ML approaches, feature engineering techniques, evaluation datasets, and performance metrics used for these countermeasures. This review also aims at identifying gaps in research on ML-based data exfiltration countermeasures. Method: We used a Systematic Literature Review (SLR) method to select and review {92} papers. Results: The review has enabled us to (a) classify the ML approaches used in the countermeasures into data-driven, and behaviour-driven approaches, (b) categorize features into six types: behavioural, content-based, statistical, syntactical, spatial and temporal, (c) classify the evaluation datasets into simulated, synthesized, and real datasets and (d) identify 11 performance measures used by these studies. Conclusion: We conclude that: (i) the integration of data-driven and behaviour-driven approaches should be explored; (ii) There is a need of developing high quality and large size evaluation datasets; (iii) Incremental ML model training should be incorporated in countermeasures; (iv) resilience to adversarial learning should be considered and explored during the development of countermeasures to avoid poisoning attacks; and (v) the use of automated feature engineering should be encouraged for efficiently detecting data exfiltration attacks.

Bushra Sabir, M. Ali Babar, Raj Gaire et al.

ML-based Phishing URL (MLPU) detectors serve as the first level of defence to protect users and organisations from being victims of phishing attacks. Lately, few studies have launched successful adversarial attacks against specific MLPU detectors raising questions about their practical reliability and usage. Nevertheless, the robustness of these systems has not been extensively investigated. Therefore, the security vulnerabilities of these systems, in general, remain primarily unknown which calls for testing the robustness of these systems. In this article, we have proposed a methodology to investigate the reliability and robustness of 50 representative state-of-the-art MLPU models. Firstly, we have proposed a cost-effective Adversarial URL generator URLBUG that created an Adversarial URL dataset. Subsequently, we reproduced 50 MLPU (traditional ML and Deep learning) systems and recorded their baseline performance. Lastly, we tested the considered MLPU systems on Adversarial Dataset and analyzed their robustness and reliability using box plots and heat maps. Our results showed that the generated adversarial URLs have valid syntax and can be registered at a median annual price of \$11.99. Out of 13\% of the already registered adversarial URLs, 63.94\% were used for malicious purposes. Moreover, the considered MLPU models Matthew Correlation Coefficient (MCC) dropped from a median 0.92 to 0.02 when tested against $Adv_\mathrm{data}$, indicating that the baseline MLPU models are unreliable in their current form. Further, our findings identified several security vulnerabilities of these systems and provided future directions for researchers to design dependable and secure MLPU systems.