Sana Mourad

Ms. Sana Mourad

orcid

Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
19 August 2026 in Electronics
Verified
Journal Article
Open access
PDF file
SMSana Mourad
Sana Mourad
EAEmad Eddien Awad Abdallah
Emad Eddien Awad Abdallah
MAMohammad Ababneh
Mohammad Ababneh
Current language model deployments face growing security challenges from prompt-based attacks, including jailbreaks, direct and indirect prompt injection, and instruction hijacking, which often evade traditional rule-based safeguards. As these models are increasingly integrated into agent-based systems, retrieval pipelines, and tool-driven workflows, such attacks exploit their natural language interfaces to bypass safety constraints and manipulate system behavior, in some cases leading to data leakage or unauthorized actions. Recent research has proposed a wide range of defense mechanisms ranging from prompt-level filtering and model-level detection to pipeline wrappers and multi-agent protection frameworks. Many of these methods report strong results in controlled experiments, yet their effectiveness depends heavily on the assumed threat model, the datasets used, and the evaluation protocol. This paper presents a systematic review and structured descriptive synthesis of research on defenses against prompt-based attacks in language model and agent systems. By integrating and comparing findings across multiple studies, we identify major attack categories, commonly adopted defense strategies, deployment stages, and evaluation trends, while also highlighting limitations related to generalization, robustness, and real-world applicability. The analysis reveals trade-offs between security effectiveness, performance, and system complexity as well as major gaps in benchmarks, indirect attack coverage, and multi-agent evaluation. The systematic review concludes by outlining future research priorities, including pipeline-aware defense design, adaptive and layered protection mechanisms, and more realistic evaluation practices to support the development of robust and deployable prompt security solutions.

24 March 2026 in IEEE Access
Verified
Journal Article
Open access
SMSana Mourad
Sana Mourad
MAMohammad Alnabhan
Mohammad Alnabhan
Intrusion Detection Systems (IDSs) are widely used to monitor and protect computer networks, but they often rely on very sensitive traffic data. Handling this kind of data may expose private information or even cause breaches if not treated carefully. To deal with this issue, this work introduces an adaptive hybrid obfuscation model designed to protect data privacy while keeping good detection accuracy for Machine Learning (ML)-based IDS. The model includes three different obfuscation levels—mild, moderate, and strong—each mixing hashing, noise, binning, and masking methods to change the data at different strengths. The researchers tested the framework using the TII-SSRC-23 dataset, after balancing it through under sampling and synthetic oversampling to make the classes balanced. The evaluation was done with TabNet as the main classifier and compared with other models like XGBoost, LightGBM, and k-Nearest Neighbors to verify robustness. Results showed that moderate obfuscation gives the best trade-off between privacy and performance, keeping accuracy around 85%, while strong obfuscation lowered accuracy to about 52%. Overall, the findings show that the proposed method can preserve data privacy without fully breaking model performance, giving a practical step toward safer and more privacy-aware IDS designs.