The European Union’s Artificial Intelligence Act changes the governance environment in which natural language processing (NLP) systems are developed and deployed. Its central lesson for Responsible NLP is that responsibility does not follow automatically from a model’s architecture or benchmark score. It is distributed across three layers: a general-purpose model, the task-specific system built around it, and the institution in which that system acts. This piece analyses those layers through recruitment, education, public-service chatbots, public-interest text generation, and general-purpose language models. It shows how the Act turns principles into duties concerning risk management, data governance, documentation, transparency, human oversight, and robustness, together with complaint, explanation, and procedural-accountability mechanisms, while also examining its interaction with data-protection law. The analysis reveals a deeper research challenge: compliance requires an evidence chain connecting intended purpose, affected people, evaluation, institutional workflow, and post-deployment monitoring. Regulation cannot create that evidence by itself. Multilingual fairness, construct validity, meaningful explanation, uncertainty communication, and effective contestability remain scientific and sociotechnical problems. The EU AI Act should therefore be understood as a legal floor for Responsible NLP, not its scientific ceiling.

Natural language processing (NLP) has moved from a specialized research field into the everyday infrastructure of writing, search, translation, education, journalism, public administration, and scientific work. This transition changes what counts as progress. Accuracy, fluency, and benchmark performance remain important, but they are no longer sufficient when language technologies shape knowledge, decisions, identities, and public trust. This column introduces Responsible NLP as a research orientation that integrates fairness, transparency, privacy, safety, cultural diversity, environmental awareness, and human agency across the full life cycle of language technologies. It argues that responsibility is not an external constraint on innovation, but a condition for meaningful and trustworthy innovation. Future research must therefore ask not only whether an NLP system works but also for whom it works, under which assumptions, with what risks, and with what forms of accountability.

Online counselling services have seen increased use in recent years, providing critical emergency mental health support. These interactions are typically long, complex, and varied in the dialogue between help seekers and counsellors. The lack of domain-specific models, especially in low-resource languages, poses a significant challenge for the automatic detection of suicide risk in online chat services for mental health support. To address this challenge, our approach adapts a general-purpose large language model (LLM) to the suicide prediction task that employs a two-stage classification architecture to deal with sparse and imbalanced data. It extends the state of the art by: (1) incorporating psychological theory into model training and (2) capturing key aspects of conversation structure in counselling sessions. We evaluate the performance of the proposed LLM against the state-of-the-art LLMs for suicide detection on thousands of conversations in the Hebrew language from a leading national online counselling service in Israel. Results show that the proposed LLM outperformed existing state-of-the-art approaches in detecting suicide risk, as measured by relevant literature metrics. Moreover, the LLM outperforms other approaches even in the early stages of a conversation, which is crucial for real-time detection in practice. We also discuss the ethical implications of combining LLMs in counselling services. The contributions of this work are (1) extending existing LLM architectures to incorporate domain-specific information; (2) evaluating LLM technologies in the context of socially relevant problems; and (3) introducing novel LLM tools for resource-constrained languages.

Neural Machine Translation (NMT), a subfield of Natural Language Processing, has seen significant advancements with the emergence of transformer architectures and generative artificial intelligence, demonstrating remarkable performance in various languages. However, translating Arabic dialects remains a notable challenge that becomes very pronounced primarily due to their morphological complexity and divergence from standardised grammatical rules. In this paper, we present a hybrid approach for translating the Maghrebi dialects into/from Modern Standard Arabic (MSA). The approach takes advantage of the strengths of the transformer architecture and the BERT language model for transfer learning of representations. To achieve this, we incorporated BERT embeddings into the encoder and decoder stacks of the transformer architecture. The BERT architecture, which we utilised, was trained in a self-supervised manner on Maghrebi dialects and Arabic corpora. The resulting BLEU/BERTScore/ChrF/METEOR scores for the approach were 14.148/79.414/28.885/28.428 and 8.961/20.994/19.465 (BLEU/ChrF/METEOR) for the translation in both directions using the raw data, demonstrating competitive performance compared to ChatGPT and Gemini Large Language Models (LLMs). Furthermore, we evaluated the approach using an ablation study with fine-tuned NLLB-200 and against three combinations of tokeniser techniques used in conjunction with the transformer architecture: Byte-Pair Encoding (BPE) tokeniser, WordPiece tokeniser, and BERT tokeniser. Both evaluations, including human evaluation, confirm the efficacy of our method.

Over the next decade, I expect more and more computing on phones and small computers and less on clouds and supercomputers. Supercomputers are super impressive to engineers, but not to economists because economies of scale have more to do with the size of the market than the size of the machine. Clouds are like wire-wrapped Cray computers; they were never designed for mass production. There was never much supply or demand. Supercomputers and clouds are too expensive and burn too much power. The future is more promising for phones because the market is larger. In addition, there are a number of other advantages to computing on phones and small (commodity) machines designed for mass markets: privacy, power, size, weight, latency, bandwidth, and especially affordability.

Natural language processing (NLP) technologies increasingly shape public life, yet their deployment for social good remains unevenly distributed across domains, languages, and geographies. This piece inaugurates the NLP for Social Good column in this journal. In this piece, I map the current state of NLP for Social Good (NLP4SG) across nine application domains. The picture that emerges is one of striking imbalance: AI harms, inclusion, and digital violence attract the bulk of research attention, while poverty, peacebuilding, and environmental protection remain critically underexplored. I argue that the field must address three structural gaps, domain coverage, linguistic diversity, and evaluation methodology, if NLP is to fulfil its potential as a force for equitable social progress. The piece concludes with five directions that I believe will define the next chapter of NLP4SG research.

There have been a number of disturbing posts on social media about large conferences. The good news is that we are past denial. There is agreement that reviewing is broken. Many posts mention toxicity. As a founder of EMNLP, it hurts to admit that there is so much toxicity that I can no longer support the process. I will continue to review for journals and smaller venues, but not for larger conferences until better ways are found to manage the crowds. Enough negativity. What can we do to be part of the solution? Many constructive suggestions have been proposed elsewhere and will be discussed here. Scale is the primary requirement. Traditional processes do not scale with exponential growth. Many attempts to fix the process (like ARR) make a bad situation worse by introducing additional bloat, while failing to address the primary requirement: scale. Process simplification will help. Steps should be removed unless they address the scaling challenge: (a) too many submissions and (b) too few qualified reviewers. A nomination process is proposed to address both (a) and (b).