When multiple normative frameworks yield equally valid answers, which framework should a language model select, and can it answer correctly within it? We formalize this as normative pluralism and study it in Islamic finance with a four-choice taxonomy that decomposes framework selection from within-framework correctness. The decomposition exposes a failure mode invisible to two-choice instruments, the stereotype trap: a model leans toward the culturally appropriate framework but produces factually wrong answers within it. Across twelve models, two languages, and fifty demographic signals, cultural cues redirect framework selection at severe cost: redirection and correctness are inversely coupled (r=−0.89 Arabic, r=−0.81 English). Under the strongest signal, large open-weight models reach 97% Islamic-frame selection, a result two-choice instruments would report as near-perfect alignment, yet 57-66% of those responses misquote the Islamic rule. We argue that cultural default bias is competence-conditioned routing: models default to the framework where they are most accurate, and the cues prescribed as mitigation expose the competence gap they were meant to correct.

Jais 2 is a family of open-weight, Arabic-centric large language models designed for strong Arabic performance and broad cultural coverage. The family includes, to our knowledge, the largest open Arabic-centric LLM trained from scratch, with 70B parameters, alongside an efficient 8B model. A custom Arabic-centric vocabulary supports efficient training and inference, while an optimized architecture and training recipe improve compute efficiency. Jais 2 achieves the highest scores on the Open Arabic LLM Leaderboard v2 (OALL2) and AraGen v1 among the evaluated open models, while remaining competitive on English benchmarks. It also obtains strong results in Arabic poetry, dream interpretation, translation, and summarization. On the IslamicQA domain, Jais 2-70B achieves the second-highest overall average among the evaluated large open models, while attaining the highest scores on PalmX and the held-out in-house IslamicQA benchmark. We release both model sizes as open-weights on Hugging Face under a commercially permissive license. Jais 2-70B is also available as a chat application on the Web, iOS, and Android, reaching inference speeds of up to 2,000 tokens per second on Cerebras hardware.

Multimodal retrieval-augmented generation (RAG) is usually evaluated with clean evidence, although deployed systems may retrieve topically relevant but unreliable text or images. We introduce QIMG-7, a controlled post-retrieval stress test for multi-sentence factual QA comprising 110 questions from four datasets, seven visual-evidence pollution families, and 16 paired clean/polluted regimes per question (1,760 rows per method). Across four generator/gate stacks, Full-MM is fragile under polluted text: in the main gpt-4o-mini stack, support decreases from 0.908 with clean text to 0.490 with polluted text. We propose source-aware trust resolution (SATR), a training-free family that compares Parametric, Text-only, and Full-MM candidate answers and selectively chooses a branch or falls back. Field-Selector obtains the highest numerical balanced score, 0.816, improving over Full-MM by 11.7 points and the Cascaded Router by 2.2 points. TextTrust, an image-blind text-reliability baseline, reaches 0.800; Field-Selector’s 1.5-point balanced advantage over it is not statistically resolved (95% CI [-0.006, 0.036]). In this text-first setting, the results support selective trust rather than unconditional multimodal fusion. Artifacts are available at https://github.com/mbzuai-nlp/Trust_Before_Fusion.

Dark humor often relies on subtle cultural nuances and implicit cues that require contextual reasoning to interpret, posing safety challenges that current static benchmarks fail to capture. To address this, we introduce a novel multimodal, multilingual benchmark for detecting and understanding harmful and offensive humor. Our manually curated dataset comprises 3,000 texts and 6,000 images in English and Arabic, along with 1,200 videos spanning English, Arabic, and language-independent (universal) contexts. Unlike standard toxicity datasets, we enforce a strict annotation guideline: distinguishing Safe jokes from Harmful ones, with the latter further classified into Explicit (overt) and Implicit (covert) categories to probe deep reasoning. We systematically evaluate state-of-the-art (SOTA) open- and closed-source models across all modalities. Our findings reveal that closed-source models significantly outperform open-source models, with a notable performance difference between English and Arabic in both, underscoring the critical need for culturally grounded, reasoning-aware safety alignment. The code and dataset are available at https://github.com/insait-institute/harm-or-humor.

Content warning: This paper contains offensive text and meme imagery, video clips, audio, and other audiovisual material that may include sexual, violent, discriminatory, religious, disability-related, or historically sensitive material. Reader discretion is advised.

Prior studies have shown that distinguishing text generated by Large Language Models (LLMs) from human-written one is highly challenging for humans, and often no better than random guessing. To verify the generalizability of this finding across languages and domains, we perform an extensive case study to identify the upper bound of human detection accuracy. Across 16 datasets covering 9 languages and 9 domains, 19 annotators achieved an average detection accuracy of 87.6%, thus challenging previous conclusions. We find that major gaps between human and machine text lie in concreteness, cultural nuances, and diversity. Prompting by explicitly explaining the distinctions in the prompts can partially bridge the gaps in over 50% of the cases. However, we also find that humans do not always prefer human-written text, particularly when they cannot clearly identify its source. We release our dataset, the human labels, and the annotator metadata at https://github.com/xnlp-lab/HumanEval-MGT.

The ease of access to large language models (LLMs) has enabled a widespread of machine-generated texts, and now it is often hard to tell whether a piece of text was human-written or machine-generated. This raises concerns about potential misuse, particularly within educational and academic domains. Thus, it is important to develop practical systems that can automate the process. Here, we present one such system, LLM-DetectAIve, designed for fine-grained detection. Unlike most previous work on machine-generated text detection, which focused on binary classification, LLM-DetectAIve supports four categories: (i) human-written, (ii) machine-generated, (iii) machine-written, then machine-humanized, and (iv) human-written, then machine-polished. Category (iii) aims to detect attempts to obfuscate the fact that a text was machine-generated, while category (iv) looks for cases where the LLM was used to polish a human-written text, which is typically acceptable in academic writing, but not in education. Our experiments show that LLM-DetectAIve can effectively identify the above four categories, which makes it a potentially useful tool in education, academia, and other domains.LLM-DetectAIve is publicly accessible at https://github.com/mbzuai-nlp/LLM-DetectAIve. The video describing our system is available at https://youtu.be/E8eT_bE7k8c.