Natural language processing research from GCC institutions, covering Arabic NLP, multilingual models, text classification, named entity recognition, and machine translation.
The International Affairs Office at the UAE Presidential Court has officially launched 'Zayed', an AI-powered spokesperson. This digital entity is designed to represent the office in various international engagements and communications, leveraging advanced artificial intelligence capabilities. The initiative aims to enhance the UAE's digital diplomacy efforts and integrate AI into high-level governmental communications. Why it matters: This development signifies the UAE's proactive integration of advanced AI into its governmental and diplomatic operations, setting a precedent for digital public communication in the Middle East.
A study investigated the impact of domain-specific pretraining on Transformer models for modeling digital pragmatics in Arabic-English code-switched discourse. It compared MARBERT and XLM-R, using BERT as a baseline, on a dataset of 11695 X posts. MARBERT significantly outperformed XLM-R, achieving a Macro F1 of 0.85 on an independent test set compared to XLM-R's 0.52. The research concludes that Transformer performance in specialized pragmatic classification tasks relies more on a domain-specific pretraining profile than on multilingual coverage alone. Why it matters: This highlights the critical importance of tailored pretraining for achieving optimal performance in complex Arabic NLP tasks, particularly in code-switching contexts prevalent in the region.
Researchers introduced YallaMorph, a new large-scale benchmark designed to evaluate Arabic morphological generation capabilities in large language models (LLMs). The benchmark covers over 600,000 entries across verbs, nouns, adjectives, their cliticized forms, and invalid configurations. Evaluations on multilingual and Arabic-oriented LLMs demonstrated that Arabic morphological generation remains challenging, particularly for cliticized, unseen, and morphologically rare forms. Why it matters: This benchmark provides a critical tool for advancing the development and accuracy of Arabic LLMs by directly addressing a core linguistic challenge for the language.
Researchers introduced EDRAC, the first large-scale benchmark for dialectal Arabic machine reading comprehension (MRC) and generative question answering (QA). EDRAC covers five major dialects: Egyptian, Moroccan, Emirati, Syrian, and Saudi Arabic, comprising 499 passages and 4,977 QA pairs generated through a human-LLM collaborative pipeline. Benchmarking Arabic-centric and multilingual LLMs on EDRAC revealed significant discrepancies between semantic answer quality and dialectal fidelity, indicating limitations of current evaluation metrics. Why it matters: This benchmark addresses a critical resource gap in dialectal Arabic NLP, offering a challenging tool for developing and evaluating models capable of understanding and generating diverse regional Arabic variants.
Researchers introduce AraDetox, a new multi-dialect Arabic detoxification dataset containing 10,500 harmful social-media posts and 84,000 detoxified rewrites. The rewrites were generated using GPT-5 and Gemini 2.5 Flash, covering Modern Standard Arabic, Gulf, Levantine, and Egyptian Arabic. Human and automatic evaluations confirmed successful harmful language removal, semantic preservation, and dialectal alignment. The dataset is publicly available to support future research in Arabic detoxification and safe text generation. Why it matters: This dataset addresses the underexplored area of Arabic text detoxification, providing a large-scale, multi-dialect resource critical for developing more ethical and robust Arabic NLP applications.
A comprehensive bibliometric and topic-based study analyzed 7,120 Arabic NLP papers published between 1960 and 2026, sourced from multiple academic platforms, using BERTopic for topic modeling, regression, and network analysis. The study found a significant publication surge after 2020, driven by transformer models and LLMs, identifying 19 key themes in Arabic NLP research. Saudi Arabia, the United States, and Egypt lead in research output, with the analysis also highlighting understudied areas like summarization for Maghrebi, Iraqi, and Sudanese dialects. Why it matters: This analysis provides a crucial quantitative overview of Arabic NLP research trends, identifying significant gaps and offering recommendations to guide future research, particularly in under-resourced dialects and culturally aligned benchmarks.
A systematic study investigated whether fine-tuning Large Language Models (LLMs) on cultural data improves figurative language understanding and vice versa. Researchers used four models, including ALLaM-7B and Fanar-1-9B, and six Arabic datasets covering cultural commonsense, proverbs, and poetry across various dialects. They found that fine-tuning on poetry improved idiom comprehension by 2.33%, suggesting a transfer of non-literal meaning understanding across figurative types. Why it matters: This research highlights the complex challenges of integrating nuanced cultural and figurative language understanding into LLMs, particularly for Arabic, suggesting that fine-tuning alone may not fully capture these intricate relationships.
HalluTruthQA-4K is an expanded corpus comprising 4,000 expert-curated Arabic question-answering instances designed for fine-grained hallucination detection and truth verification in large language models. The resource includes model-generated responses, verified reference answers, distractors, character-level erroneous spans, human explanations, and hierarchical hallucination types across four knowledge domains. It features 1,643 hallucinated and 2,357 non-hallucinated responses, with 1,843 annotated erroneous spans, and serves as the official dataset for Track 2 of the HalluScoring 2026 shared task. Why it matters: This comprehensive, expert-annotated corpus provides a crucial foundation for advancing research and development in building more reliable and factually accurate Arabic large language models.
Researchers introduced HalluTruthQA-4K, an expanded corpus comprising 4,000 expert-curated Arabic question-answering instances designed for hallucination detection and truth verification. This resource spans four knowledge-intensive domains: Islamic knowledge, history, science, and geography, and serves as the official dataset for Track 2 of the HalluScoring 2026 shared task. For hallucinated responses, the corpus provides character-level erroneous spans, human-written explanations, and hierarchical hallucination types, alongside verified reference answers and distractors. Why it matters: HalluTruthQA-4K provides a crucial fine-grained resource for evaluating and improving the factual reliability and trustworthiness of Arabic large language models.
Researchers investigated the functional necessity of visual distinctions in Arabic script for NLP by comparing standard dotted, dotless, and arbitrarily remapped Arabic. They generated 2,000 random character remappings constrained to 19 undotted rasms, evaluating them across tasks like language modeling, text classification, and machine translation. The study found that neither preserving original character distinctions nor traditional rasm-based groupings is necessary for strong NLP performance, with random remappings achieving competitive results while reducing vocabulary size and training costs. Why it matters: These findings suggest that Arabic NLP models primarily rely on stable distributional structure rather than visual iconicity, potentially leading to more efficient and effective Arabic language processing.
Researchers introduced AraBehave, a benchmark comprising 1,623 culturally grounded, open-ended Arabic prompts and over 29,000 human judgments to evaluate the cultural appropriateness of large language models (LLMs). The study found that cultural appropriateness involves two distinct components: normative stance and grounded cultural accuracy, with general-purpose models often failing on stance and Arabic-centric models on accuracy. Providing cultural instructions significantly improved the normative stance of general-purpose models, while grounding correlated with model scale and Arabic alignment data. Why it matters: This benchmark provides a critical tool for assessing how LLMs behave in nuanced Arabic cultural contexts, moving beyond mere factual knowledge to address ethical and social acceptance, which is crucial for their responsible deployment and trustworthiness in the region.
This study evaluated the adversarial robustness of five state-of-the-art Arabic Language Models against various Arabic adversarial attacks at character, word, and sentence levels. It found that diacritic insertion could reduce model accuracy by up to 92%, while manipulating Arabic conjunctions led to a 58% accuracy degradation, and paraphrasing reduced performance by an average of 76%. While adversarial training improved overall resilience, particularly for MARBERT and AraBERT, challenges against character-level noise persist. Why it matters: These findings are crucial for understanding and mitigating security vulnerabilities in Arabic AI, guiding the development of more robust and safe Arabic NLP systems.
Researchers introduce HalluTruthQA, a new fine-grained benchmark designed for hallucination detection, localization, and explanation in Arabic question answering. This benchmark comprises 2,400 expert-curated examples across Islamic knowledge, history, science, and geography, featuring character-level error spans, human explanations, and various hallucination types. The study evaluated four open-source Arabic LLMs (ALLaM-7B, Falcon-H1R-7B, Qwen3-32B, SILMA) across detection, localization, factual verification, and explanation tasks, revealing no single model outperforms others across all metrics. Why it matters: HalluTruthQA provides a critical tool for advancing the factual accuracy and reliability of Arabic LLMs by enabling more granular and comprehensive hallucination evaluation beyond response-level detection.
Researchers have introduced HalluTruthQA, a new fine-grained benchmark designed for hallucination detection, localization, and explanation in Arabic Question Answering. The benchmark comprises 2,400 expert-curated examples spanning four knowledge-intensive domains: Islamic knowledge, history, science, and geography, with detailed annotations including character-level erroneous spans and human-written explanations. Four open-source LLMs ( extsc{Allam}, extsc{Falcon-H1}, extsc{Qwen32}, and extsc{Silma}) were evaluated, demonstrating varied performance across detection, localization, factual verification, and explanation tasks. Why it matters: This benchmark offers a comprehensive tool for evaluating and enhancing the factual accuracy and trustworthiness of Arabic LLMs, promoting more sophisticated assessment beyond simple hallucination detection.
Researchers introduced CAMMAR (Culture-Aware Matryoshka for Metaphorical Arabic Representations), a framework designed to organize meaning in Arabic language models into nested lexical, cultural, and metaphorical embedding subspaces, addressing the issue of "semantic smearing." The framework, inspired by Al-Jurjani's theory of nazum, provides a training-free geometric measure of metaphoricity. Evaluated on a new span-annotated Arabic metaphor dataset, CAMMAR achieved an AUC of up to 0.84, effectively detecting metaphor when inter-layer geometry was shaped by paired supervision. Why it matters: This research offers a novel approach to enhancing the cultural and metaphorical understanding of Arabic AI, potentially leading to more nuanced and accurate Arabic language models.
This study investigates methods to steer Arabic Large Language Models (LLMs) towards generating specific dialects, addressing the challenge of data scarcity for dialectal Arabic. Researchers identified sparse neuron populations encoding dialect-specific features and developed a vector-steering approach using dialect-specific activation directions. These inference-time methods allow for controlling dialectal output by amplifying or suppressing neuron activity or injecting specific vectors. Why it matters: This research offers a principled, interpretability-grounded framework to improve dialectal accuracy in Arabic LLMs without fine-tuning, crucial for enhancing their utility in the diverse Arabic-speaking world.
A new paper evaluated twelve Small Language Models (SLMs) on Arabic natural language processing tasks, utilizing a benchmark of 240 Arabic test items across eight domains and ten language skills. The models were assessed in a zero-shot setting, with responses scored using a multi-model LLM-as-a-judge framework involving GPT-4.1 Mini, Claude Haiku 4.5, and DeepSeek-Chat. Gemma 3 (12B) achieved the highest overall score (4.548/5), followed by Aya and C4AI Command Arabic, with results suggesting that strong Arabic alignment and instruction-following are crucial for performance. Why it matters: This benchmark offers a standardized method for evaluating compact Arabic language models, guiding future development towards more efficient, reliable, and culturally relevant Arabic AI systems.
The UAE Presidential Court has unveiled "Zayed," an AI-powered spokesperson designed to represent the court. This new digital entity is expected to deliver official announcements and engage with the public, leveraging advanced artificial intelligence technologies. The initiative represents a significant step in the UAE's efforts to integrate AI into government communications and public services. Why it matters: This deployment showcases the UAE's proactive adoption of AI in high-profile public sector roles, potentially setting a precedent for AI integration in government communication across the region.
This paper presents a methodology for digitizing and encoding the Al-Mawrid Arabic-English dictionary using the ISO Lexical Markup Framework (LMF) and TEI Lex-0 guidelines. The research resolves structural ambiguities and inconsistencies, achieving a structural parsing accuracy of 91% and high precision/recall for information extraction, such as 85% precision for synonyms. It also discusses limitations of TEI Lex-0 for Arabic phenomena and explores Linguistic Linked Open Data (LLOD) integration. Why it matters: This work provides a crucial, standardized computational lexicon for Arabic, addressing a significant gap in Arabic lexical infrastructure and offering a reproducible workflow for retro-digitization efforts in Arabic NLP and Digital Humanities.
This paper presents a methodology for digitizing and encoding the Al-Mawrid Arabic-English dictionary, transforming it into a standardized computational lexicon using the ISO Lexical Markup Framework (LMF) and TEI Lex-0 guidelines. The research, based on an empirical analysis of the letter Ayn (4.6% of the dictionary), achieved a structural parsing accuracy of 91%. Quantitative evaluation showed high performance for information extraction rules, including 85% precision and 98% recall for synonyms. Why it matters: This work addresses a significant gap in Arabic lexical infrastructure, providing an interoperable, machine-tractable resource and a reproducible workflow for retro-digitizing complex legacy bilingual lexicons for Arabic NLP and Digital Humanities.
Researchers have introduced BloomBench, a new cognitively human-grounded, bilingual (English-Arabic) multimodal benchmark for Vision-Language Models (VLMs), as part of the Almieyar benchmarking series. Grounded in Bloom's Taxonomy, it systematically evaluates six levels of cognition—Remember, Understand, Apply, Analyze, Evaluate, Create—through carefully designed image-question-answer tasks. A comprehensive study using BloomBench revealed that state-of-the-art VLMs exhibit strong semantic understanding but struggle significantly with factual recall and creative synthesis, alongside a critical performance gap between Arabic and English. Why it matters: This benchmark provides a crucial tool for diagnosing cognitive weaknesses in current VLMs and lays the groundwork for developing more cognitively aligned and inclusive multimodal AI, particularly for cross-lingual applications.
Researchers proposed a four-stage NLP framework combining schema-constrained LLM extraction, Sentence-BERT (SBERT) alignment with ESCO, an adjudication protocol, and a verification mechanism for curriculum-labor market alignment. The framework was instantiated for the ABET-accredited BSc Computer Science program at the United Arab Emirates University (UAEU), extracting 400 competency records from the study plan and aligning them with 30 job postings. The extractor achieved a Cohen's kappa of 0.79 on the skill slot and surfaced interpretable supply-demand gaps in general, transversal, algorithms, and software engineering skills, with a minimal gap in AI and data science. Why it matters: This framework provides a robust, NLP-driven method to identify crucial skill gaps in higher education curricula, directly supporting quality assurance and workforce development initiatives in the region.
The UAE has spotlighted Agentic AI as a key element for the future of government communication. Concurrently, the government launched the Government Media Content Guideline to regulate content in this evolving landscape. This initiative underscores the UAE's strategic move to integrate advanced AI technologies into its public sector operations. Why it matters: This development signifies a proactive governmental approach to AI adoption and regulation, potentially setting a precedent for other nations in the Middle East in managing AI-powered public communication.
Researchers from MBZUAI have proposed a new taxonomy of eight temporal frames and studied their persuasive use in news discourse. They created a multilingual dataset by expertly annotating 458 English and German news articles, identifying over 2,000 temporally framed sentences and approximately 3,000 annotations. Their experiments demonstrated that temporal framing is learnable at the sentence level, with supervised models significantly outperforming zero-shot classification approaches. Why it matters: This research provides a valuable dataset and methodology for understanding how time-related language shapes interpretation in news, contributing to advancements in NLP for media analysis and potentially countering disinformation.
ArabDiscrim is a new corpus comprising 293,000 public Arabic Facebook posts from 2014 to 2024, specifically curated to discuss racism and discrimination. Unlike prior Twitter-centric datasets, it incorporates platform-native engagement signals, 200 curated terms with morphological regex families, and 20 discrimination axes. The resource also provides explicit attribution patterns and is released under a restricted research-use license for ethical compliance. Why it matters: This dataset provides a unique, ecologically valid foundation for fairness-oriented and platform-aware Arabic Natural Language Processing, moving beyond existing Twitter-centric resources.
This paper reflects on two decades of building NLP resources and research infrastructure for Arabic, an historically underserved language. The first decade focused on foundational linguistic infrastructure, while the second shifted towards computational social science and socially oriented applications. The authors highlight three lessons: dataset building is a social process, communities often matter more than shared tasks, and computational social science exposes challenges beyond traditional NLP training. Why it matters: The paper argues that the most difficult problems in developing NLP for underserved communities are social, institutional, and epistemic, offering critical insights for future research directions in Arabic AI.
Researchers have introduced JobArabi, a new large-scale corpus consisting of 20,528 Arabic job announcements collected from X between January 2024 and October 2025. The dataset was compiled using a linguistically informed query framework covering various Arabic recruitment expressions, offering metadata like timestamps and geolocation for detailed analysis. Quantitative analysis of JobArabi reveals sociolinguistic patterns, including persistent gendered hiring language, regional occupational demand variations, and emotional framing in recruitment messages. Why it matters: This corpus provides a valuable resource for research in Arabic NLP, computational social science, and digital labor studies, offering unique insights into labor market communication and linguistic change in the Arab world.
Researchers developed an Arabic NLP framework designed for large-scale financial sentiment analysis specifically tailored to the Saudi market. The framework integrates official financial news and social media, constructing an 84K-sample Arabic financial corpus through a multi-stage pipeline encompassing data collection, cleaning, and sentiment annotation. It employs Transformer-based NER and a curated company lexicon to link textual mentions to canonical company identifiers, assigning five-class sentiment labels for analyzing sentiment dynamics relative to stock market behavior on the Saudi Exchange. Why it matters: This research addresses a critical gap in Arabic financial NLP resources, offering a scalable method to understand investor sentiment in a key Middle Eastern market.
Researchers at NYU Abu Dhabi have developed an AI system capable of translating spoken language into sign language. This innovative technology aims to enhance communication accessibility for individuals who are deaf or hard-of-hearing. The system leverages advancements in artificial intelligence, likely combining natural language processing for speech understanding and computer vision for sign generation. Why it matters: This development has the potential to significantly improve inclusion and communication for deaf communities within the Middle East and globally, bridging critical communication gaps.
Researchers have proposed the Cylindrical Representation Hypothesis (CRH) to address the instability and unpredictability observed in steering large language models, an issue not fully explained by the existing Linear Representation Hypothesis (LRH). CRH suggests that overlapping concept contributions lead to a sample-specific axis-orthogonal structure, comprising a central axis for concept generation and a surrounding normal plane for steering sensitivity. This framework identifies intrinsic uncertainty at the 'sensitive sector' level within the plane, providing a principled explanation for fluctuations in steering outcomes. Experiments verify the existence of this cylindrical structure and demonstrate CRH's practical utility in interpreting real-world model steering behavior, with code available on GitHub from mbzuai-nlp. Why it matters: This research from MBZUAI offers a crucial theoretical advancement in understanding and potentially improving the control and reliability of large language models.
Researchers from MBZUAI have proposed the Cylindrical Representation Hypothesis (CRH) to explain the instability and unpredictability observed in large language model steering. CRH relaxes the orthogonality assumption of the existing Linear Representation Hypothesis, positing a cylindrical structure where a central axis captures concept differences and a surrounding normal plane controls steering sensitivity. The hypothesis suggests that the intrinsic uncertainty in identifying specific sensitive sectors within this normal plane accounts for why steering outcomes frequently fluctuate even with well-aligned directions. Why it matters: This research offers a more robust theoretical framework for understanding and potentially improving the control and reliability of large language models.
Researchers at MBZUAI have developed a new method for controllable poetry generation in Arabic and its dialects, moving beyond traditional analysis tasks for Arabic poetry within Large Language Models (LLMs). They introduce a large-scale, instruction-based dataset in Modern Standard Arabic (MSA) and various Arabic dialects, enabling LLMs to perform tasks like writing, revising, and continuing poems based on user criteria. Experiments show that fine-tuning LLMs on this dataset results in models capable of generating poetry aligned with user requirements, validated by automated metrics and human evaluation. Why it matters: This work represents a significant advancement in Arabic Natural Language Processing, offering tools for creative expression and cultural preservation while opening new avenues for user-guided content generation in culturally rich text forms.
Google has introduced a new AI feature in the United Arab Emirates, designed to provide more intelligent and personalized answers to users. This feature reportedly leverages user data, with consent, to enhance its responsiveness and relevance. The rollout in the UAE signifies the expansion of Google's advanced AI services into the Middle East market. Why it matters: This launch represents increased access to sophisticated AI tools for consumers and businesses in the UAE, potentially accelerating AI adoption and innovation in the local digital economy.
Researchers proposed a severity-aware weighted loss method to fine-tune Arabic language models for medical text generation, prioritizing severe clinical cases. This approach utilizes soft severity probabilities, derived from an AraBERT-based classifier, to dynamically scale token-level loss contributions during optimization on the MAQA dataset. The method consistently improved performance across ten Arabic LLMs, with AraGPT2-Base increasing from 54.04% to 66.14% and AraGPT2-Medium from 59.16% to 67.18%. Why it matters: This novel fine-tuning strategy addresses a critical limitation in medical AI by enhancing the safety and reliability of Arabic medical large language models, particularly in high-stakes clinical scenarios.
Arabic-DeepSeek-R1 is an application-driven, open-source Arabic Large Language Model (LLM) that has achieved a new state-of-the-art (SOTA) across the Open Arabic LLM Leaderboard (OALL). The model utilizes a sparse Mixture-of-Experts (MoE) backbone and a four-phase Chain-of-Thought (CoT) distillation scheme, which incorporates Arabic-specific linguistic verification and regional ethical norms. It records the highest average score on the OALL suite and outperforms proprietary frontier systems like GPT-5.1 on a majority of benchmarks evaluating comprehensive Arabic language-specific tasks. Why it matters: This work offers a validated and cost-effective framework for developing high-performing, culturally-grounded AI for under-represented languages, addressing the digital equity gap.
Researchers have developed OmniScore, a family of deterministic learned metrics designed to evaluate generative text as an alternative to Large Language Models (LLMs) used as judges. OmniScore leverages small parameter models (<1B) and was trained on approximately 564,000 synthetic instances across 107 languages, then evaluated using 8,617 manually annotated instances. It approximates LLM-judge behavior while offering low latency and consistency for various evaluation settings like reference-based and source-grounded assessments in tasks like QA, translation, and summarization. Why it matters: This development provides a practical, scalable, and reproducible method for multilingual generative text evaluation, addressing key limitations of LLM-as-a-judge approaches and offering significant benefits for AI development in linguistically diverse regions.
QIMMA is introduced as a quality-assured Arabic LLM leaderboard that places systematic benchmark validation at its core. It employs a multi-model assessment pipeline combining automated LLM judgment with human review to identify and resolve quality issues in established Arabic benchmarks. The resulting evaluation suite comprises over 52,000 samples, predominantly grounded in native Arabic content, with transparent implementation via LightEval and EvalPlus. Why it matters: This initiative provides a more reliable and reproducible foundation for evaluating Arabic Large Language Models, addressing critical quality concerns in existing benchmarks.
Researchers developed a retrieval-augmented generation (RAG) framework to improve Arabic Large Language Models (LLMs) in understanding complex historical and religious texts like the Quran and Hadith. This framework grounds LLMs in the Doha Historical Dictionary of Arabic (DHDA) through hybrid retrieval and intent-based routing. The approach significantly boosted the accuracy of Arabic-native LLMs such as Fanar and ALLaM to over 85%, closing the performance gap with proprietary models like Gemini. Why it matters: This research offers a novel method for enhancing Arabic NLP capabilities for historically nuanced texts, demonstrating the value of integrating diachronic lexicographic resources into RAG systems for deeper language understanding.
The Technology Innovation Institute (TII) has launched Falcon Perception, a new 600-million-parameter multimodal AI model. This model offers competitive performance in object segmentation, dense visual understanding, and document intelligence, rivalling larger systems like Meta’s SAM3 and Alibaba’s Qwen with significantly greater efficiency. Falcon Perception unifies image and language processing in a single architecture, designed for real-world deployment in compute-constrained environments. Why it matters: This development positions the UAE among leading nations in advanced multimodal AI, which is crucial for applications in robotics, advanced manufacturing, and autonomous platforms.
The Open Arabic LLM Leaderboard (OALL) has been launched to benchmark Arabic language models, addressing the gap in resources for non-English NLP. It incorporates datasets like AlGhafa, ACVA, and translated versions of MMLU and EXAMS from the AceGPT suite. The leaderboard uses normalized log likelihood accuracy for tasks, built around HuggingFace’s LightEval framework. Why it matters: This initiative promotes research and development in Arabic NLP, serving over 380 million Arabic speakers by enhancing the evaluation and improvement of Arabic LLMs.
Technology Innovation Institute (TII) in Abu Dhabi, in collaboration with LightOn, has launched NOOR, a 10 billion parameter Arabic natural language processing (NLP) model. The model was trained on a large, high-quality cross-domain Arabic dataset including web data, books, poetry, news, and technical information. It enables applications in automated summarization, chatbots, and personalized marketing. Why it matters: NOOR represents a significant advancement in Arabic NLP, potentially enabling more sophisticated AI applications tailored to the Arabic language and regional needs.
Abu Dhabi’s Technology Innovation Institute (TII) has launched Falcon-H1 Arabic, a new large language model based on a hybrid Mamba-Transformer architecture. The Falcon-H1 family comes in 3B, 7B, and 34B parameter sizes and outperforms existing models on the Open Arabic LLM Leaderboard (OALL). The model features improvements in data quality, dialect coverage, and long-context stability. Why it matters: This release strengthens the UAE's position in Arabic AI and provides a high-performing model tailored to the linguistic and cultural needs of the region.
The paper introduces SectEval, a new benchmark to evaluate sectarian biases in LLMs concerning Sunni and Shia Islam, available in English and Hindi. Results show significant inconsistencies in LLM responses based on language, with some models favoring Shia responses in English but Sunni in Hindi. Location-based experiments further reveal that advanced models adapt their responses based on the user's claimed country, while smaller models exhibit a consistent Sunni-leaning bias.
MBZUAI researchers have developed an automatic interview system that uses LLMs to elicit nuanced, role-specific information from job candidates, improving early-stage hiring decisions. The system updates its belief about an applicant's rubric-oriented latent traits in a calibrated way based on their interview performance. Evaluation on simulated interviews showed the system's belief converges towards the simulated applicants' constructed ability levels.
The paper introduces ArabicNumBench, a benchmark for evaluating LLMs on Arabic number reading using both Eastern and Western Arabic numerals. It evaluates 71 models from 10 providers on 210 number reading tasks, using zero-shot, zero-shot CoT, few-shot, and few-shot CoT prompting strategies. The results show substantial performance variation, with few-shot CoT prompting achieving 2.8x higher accuracy than zero-shot approaches. Why it matters: The benchmark establishes baselines for Arabic number comprehension and provides guidance for model selection in production Arabic NLP systems.
The paper introduces ALPS (Arabic Linguistic & Pragmatic Suite), a diagnostic challenge set for evaluating deep semantics and pragmatics in Arabic NLP. The dataset contains 531 expert-curated questions across 15 tasks and 47 subtasks, designed to test morpho-syntactic dependencies and compositional semantics. Evaluation of 23 models, including commercial, open-source, and Arabic-native models, reveals that models struggle with fundamental morpho-syntactic dependencies, especially those reliant on diacritics. Why it matters: ALPS provides a valuable benchmark for evaluating the linguistic competence of Arabic NLP models, highlighting areas where current models fall short despite achieving high fluency.
Arabic Language Models (LMs) are primarily pretrained on Modern Standard Arabic (MSA), with an expectation of transferring to diverse Arabic dialects for real-world applications. This work explores cross-lingual transfer in Arabic LMs using probing on three Natural Language Processing (NLP) tasks and representational similarity. The findings indicate that transfer is possible but disproportionate across dialects, with some evidence of negative interference in models trained to support all Arabic dialects. Why it matters: This research highlights crucial challenges for building robust Arabic AI systems that effectively handle the significant linguistic diversity of the Arab world.
The paper introduces SalamahBench, a new benchmark for evaluating the safety of Arabic Language Models (ALMs). The benchmark comprises 8,170 prompts across 12 categories aligned with the MLCommons Safety Hazard Taxonomy. Five state-of-the-art ALMs, including Fanar 1 and 2, ALLaM 2, Falcon H1R, and Jais 2, were evaluated using the benchmark. Why it matters: The benchmark enables standardized, category-aware safety evaluation, highlighting the necessity of specialized safeguard mechanisms for robust harm mitigation in ALMs.
MBZUAI researchers introduce DuwatBench, a new benchmark for multimodal understanding of Arabic calligraphy. The dataset contains 1,272 samples across six calligraphic styles with detailed annotations to evaluate visual-text alignment. Evaluation of 13 multimodal models reveals challenges in processing calligraphic variations and artistic distortions, highlighting the need for culturally grounded AI research.
A national survey in Saudi Arabia of 330 participants reveals that 93% are actively using Generative AI, primarily for text-based tasks, while awareness and understanding remain uneven. Participants recognize benefits like productivity but caution against risks such as privacy, misinformation, and ethical misuse. The study highlights the need for AI literacy, culturally aligned solutions, and stronger frameworks for responsible deployment in Saudi Arabia.