Tarek Helmy

Building and Evaluating an Arabic Social-Media Dataset for Deepfake Text Detection

Rapid advances in large language models have enabled the creation of sophisticated deepfake text, which poses a significant threat to social media information integrity. Arabic deepfake text detection is underexplored owing to the morphological complexity of the language and its diverse regional dialects. Studies in this domain often rely on narrow datasets or focus exclusively on Modern Standard Arabic and fail to capture the informal linguistic nuances prevalent on platforms like X. This study addresses these gaps by compiling a high-quality Arabic dataset spanning six major dialects and seven distinct domains, providing a representative benchmark for real-world environments. Real Arabic texts were collected from X, deepfake texts were generated via a prompt-based approach using GPT-4o-mini across multiple deception styles, and a pretrained BERT-base-uncased model was fine-tuned for binary classification using the Hugging Face Transformers framework. Experimental results validated the high quality of the proposed dataset, The model achieved training, validation, and test accuracies of 93.7%, 93.2%, and 92.5%, respectively, with an F1 score of 92.5%. A significant finding of this study was the variation in performance across different data sources. The highest accuracy was observed in entertainment content from YouTube, suggesting that the model effectively identifies the stylistic and expressive markers inherent in synthetic conversational Arabic. Our contributions include the curation of an authentic multi-dialectal corpus and validation of transformer-based architectures tailored for identifying AI-manipulated text. This study provides a benchmark for the Arabic natural language processing community.

ArabiDeepfake: a multi-domain Arabic deepfake-text dataset and cross-domain detection benchmark

Toward practical migration to post-quantum SSH: system-level design and evaluation

The migration of remote-access and industrial communication systems from classical public-key cryptography to post-quantum cryptography (PQC) requires careful evaluation at both the protocol and system levels. This paper presents PQC-E2E-CA, a system-level evaluation framework for reviewing post-quantum and hybrid cryptographic configurations in Secure Shell (SSH). The framework integrates OQS-enabled OpenSSH and OpenSSL with Linux netem network emulation, automated experiment execution, SCP integrity verification, and statistical post-processing. The evaluation separates key exchange behavior from host key authentication. Specifically, it measures ML-KEM and hybrid ML-KEM as SSH key exchange mechanisms, and ML-DSA as a host-key signature mechanism. Experiments are conducted under controlled RTT and packet-loss conditions using a gateway virtualised client-server testbed. The results show that ML-KEM and hybrid ML-KEM can be integrated into SSH without prohibitive application-level session setup overhead in the evaluated environment. Among the evaluated configurations, ML-KEM-768 demonstrates comparatively lower SSH session establishment latency at 50 ms RTT with 0% packet loss. ML-DSA-44 achieves the lowest host-key authentication latency under the same conditions and maintains relatively stable performance at 150 ms RTT with 5% packet loss. SCP throughput results for 100 MB and 200 MB transfers indicate that sustained transfer performance is mainly influenced by RTT and transport-layer dynamics using a single dominant key exchange configuration. These findings support migration toward standardized post-quantum mechanisms in SSH-based gateway and remote-access environments, provided that algorithm choice and system configuration are validated under representative workloads and network conditions.