Safety alignment in Large Language Models (LLMs) often involves mediating internal representations to refuse harmful requests. Recent research has demonstrated that these safety mechanisms can be bypassed by ablating or removing specific representational directions within the model. In this paper, we propose the opposite approach: Rank-One Safety Injection (ROSI), a white-box method that amplifies a model’s safety alignment by permanently steering its activations toward the refusal-mediating subspace. ROSI operates as a simple, fine-tuning-free rank-one weight modification applied to all residual stream write matrices. The required safety direction can be computed from a small set of harmful and harmless instruction pairs. We show that ROSI consistently increases safety refusal rates while preserving the utility of many models across diverse architectures. Furthermore, we show that ROSI can also re-align “uncensored” models by amplifying their own latent safety directions, demonstrating its utility as an effective last-mile safety procedure. Our results show that targeted, interpretable weight steering is a cheap and effective mechanism to improve LLM safety, complementing more resource-intensive fine-tuning paradigms.

We present AraLingBench, a fully human annotated benchmark for evaluating the Arabic linguistic competence of large language mod- els (LLMs). The benchmark spans five core categories: grammar, morphology, spelling, reading comprehension, and syntax, through 150 expert designed multiple choice questions that directly assess structural language understanding. Evaluating 35 Arabic and bilingual LLMs reveals that current models demonstrate strong surface level proficiency but struggle with deeper grammatical and syntactic reasoning. AraLingBench highlights a persistent gap between high scores on knowledge-based benchmarks and true linguistic mastery, showing that many models succeed through memorization or pattern recognition rather than au- thentic comprehension. By isolating and measuring fundamental linguistic skills, AraLingBench provides a diagnostic framework for developing Arabic LLMs. The benchmark and evaluation code are available on Hugging Face and GitHub.

We present HALA, a family of Arabic-centric instruction and translation models built with our translate-and-tune pipeline. We first compress a strong AR↔EN teacher to FP8 (yielding ~2× higher throughput with no quality loss) and use it to create high-fidelity bilingual supervision. A lightweight language model LFM2–1.2B is then fine-tuned on this data and used to translate high-quality English instruction sets into Arabic, producing a million-scale corpus tailored to instruction following. We train HALA models at 350M, 700M, 1.2B, and 9B parameters, and apply slerp merging to balance Arabic specialization with base-model strengths. On Arabic-centric benchmarks, HALA achieves state-of-the-art results within both the “nano” (≤2B) and “small” (7–9B) categories, outperforming their bases. We are committed to release models, data, evaluation, and recipes to accelerate research in Arabic NLP.

Merging Large Language Models (LLMs) is a cost-effective technique for combining multiple expert LLMs into a single versatile model, retaining the expertise of the original ones. However, current approaches often overlook the importance of safety alignment during merging, leading to highly misaligned models. This work investigates the effects of model merging on alignment. We evaluate several popular model merging techniques, demonstrating that existing methods do not only transfer domain expertise but also propagate misalignment. We propose a simple two-step approach to address this problem: (i) generating synthetic safety and domain-specific data, and (ii) incorporating these generated data into the optimization process of existing data-aware model merging techniques. This allows us to treat alignment as a skill that can be maximized in the resulting merged LLM. Our experiments illustrate the effectiveness of integrating alignment-related data during merging, resulting in models that excel in both domain expertise and alignment.