Morocco

Large Language Models (LLMs) are increasingly integrated into high-stakes applications, making robust safety guarantees a central practical and commercial concern. Existing safety evaluations predominantly rely on fixed collections of harmful prompts, implicitly assuming non-adaptive adversaries and thereby overlooking realistic attack scenarios in which inputs are iteratively refined to evade safeguards. In this work, we examine the vulnerability of contemporary language models to automated, adversarial prompt refinement. We repurpose black-box prompt optimization techniques, originally designed to improve performance on benign tasks, to systematically search for safety failures. Using DSPy, we apply three such optimizers to prompts drawn from HarmfulQA and JailbreakBench, explicitly optimizing toward a continuous danger score in the range 0 to 1 provided by an independent evaluator model (GPT-5.1). Our results demonstrate a substantial reduction in effective safety safeguards, with the effects being especially pronounced for open-source small language models. For example, the average danger score of Qwen 3 8B increases from 0.09 in its baseline setting to 0.79 after optimization. These findings suggest that static benchmarks may underestimate residual risk, indicating that automated, adaptive red-teaming is a necessary component of robust safety evaluation.

EACL 2026

When Prompt Optimization Becomes Jailbreaking: Adaptive Red-Teaming of Large Language Models

llm evaluations

jailbreaking

prompt optimization

**Welcome to EMNLP 2023!**

On behalf of the EMNLP 2023 Organizing Committee, I extend a warm and heartfelt welcome to all of you to the 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP). It is with immense pleasure and excitement that we gather here in Singapore, a vibrant hub of innovation and technological advancement.

The conference program is packed with insightful presentations, thought-provoking workshops, and engaging networking opportunities. In addition to the technical sessions, we have also planned several social events that will provide you with the opportunity to connect with your colleagues.

The conference is held in person at the Resorts World Convention Centre in Singapore, and available on-line with the help of Underline.

For those in Singapore, I hope that you will find time to explore the exciting Sentosa Island, Gardens by the Bay, Singapore Botanic Gardens and the many other attractions unique to Singapore.

Of course, we are grateful to our sponsors and partners for their generous support of EMNLP 2023, whose contributions make it possible for us to host this world-class event.

Yuji Matsumoto (RIKEN AIP) 
EMNLP 2023 General Chair

To access the **EMNLP 2023** event page on Underline, you need to register for the Conference. 
Please follow **[this link](https://2023.emnlp.org/registration/)** for more details.

EMNLP 2023

EMNLP 2023 took place in Singapore from Dec 6th to Dec 10th, 2023.

technical paper

#### *Message from the General Chair, Aline Villavicencio*
I’m delighted and honoured to welcome you to the 19th Conference of the European Chapter of the Association for Computational Linguistics (EACL 2026), taking place in the beautiful city of Rabat, in Morocco, in March 24-29, 2026. EACL is the flagship European conference of the Association and EACL 2026 proudly continues our field’s tradition of excellence in scholarship, innovation, and inclusivity. I am deeply grateful to the many volunteers whose dedication, generosity, and tireless efforts have made this conference possible.
For the first time EACL is being hosted in the African continent. This is an important milestone for our community, and we are grateful to our Moroccan hosts for enabling this historic moment by bringing this edition of EACL to Rabat. We are also delighted that the Second Arabic NLP School is co-located with EACL. We hope attendees enjoy this wonderful opportunity to strengthen ties with the Computational Linguistics communities across the African continent. *[Read full message](https://drive.google.com/file/d/14NlmHvwM6fPJuMmOvVh7K0vtQbEyv3SZ/view?usp=sharing)* 

<html><button style="display: inline-flex; align-items: center; justify-content: center; white-space: nowrap; border-radius: 9999px; font-weight: bold; background: #7c3aed; color: white; font-family: 'Space Grotesk', sans-serif; height: 40px; font-size: 16px; padding: 0 20px; border: none; cursor: pointer" onclick="window.open('https://underline.io/events/522/reception','_blank')">Go to Workshops and Tutorials Program</button></html>

To access this event page, you need to log in with the **email address you registered with**. Access credentials will be sent to your email from Underline - subject line "Welcome to EACL 2026". Please be sure to check your spam email folder if you do not see an email confirmation right away.

Please log in

To access this event page, you are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://2026.eacl.org/registration/) first.

**Online Registration Form**: https://acl.swoogo.com/eacl2026

Registration Required

Welcome to the 19th Conference of the European Chapter of the Association for Computational Linguistics (EACL). EACL 2026 will be held in Rabat, Morocco, from March 24–29, 2026. 

Sign language lexicographers construct bilingual dictionaries by establishing word-to-sign mappings, where polysemous and homonymous words corresponding to different signs across contexts are often underrepresented. A usage-based approach examining how word senses map to signs can identify such novel mappings absent from current dictionaries, enriching lexicographic resources.
We address this by analyzing German and German Sign Language (Deutsche Gebärdensprache, DGS), manually annotating 1,404 word use–to–sign ID mappings 
derived from 32 words from the German Word Usage Graph (D-WUG) and 49 signs from the Digital Dictionary of German Sign Language (DW-DGS). We identify three correspondence types: Type 1 (one-to-many), Type 2 (many-to-one), and Type 3 (one-to-one), plus No Match cases. We evaluate computational methods: Exact Match (EM) and Semantic Similarity (SS) using SBERT embeddings. SS substantially outperforms EM overall 88.52\% vs. 71.31\%), with dramatic gains for Type 1 (+52.1 pp). Our work establishes the first annotated dataset for cross-modal sense correspondence and reveals which correspondence patterns are computationally tractable.
Our code and dataset are made publicly available

How Do Lexical Senses Correspond Between Spoken German and German Sign Language?

This thesis proposal addresses methodological gaps in applying NLP to social science by shifting from categorical classification to comparative scaling of grounded constructs. We first extend predictive capacity on existing specialized political datasets with prompt optimization and distillation approaches. We then develop an active learning framework for efficient comparative annotation to scale latent dimensions from large corpora. Finally, we apply this pipeline to measure benevolent sexism in Slovenian media and migration threat perception in parliamentary discourse. This work establishes a scalable workflow for moving NLP from ad-hoc classification to theoretically grounded comparative measurement.

Thesis Proposal: Measuring Prejudice at Scale

Policy-gradient reinforcement learning (RL) is widely used to improve language model reasoning, but existing methods are not compatible with diffusion language models. The primary reason for this is the difficulty of likelihood estimation with such models. We propose EMBR, a scalable off-policy framework that reformulates KL-regularized RL as an energy-based distribution matching problem. By aligning policy updates with reward signals through energy matching,EMBR avoids the overhead of on-policy learning and the variance of importance weighting. We further derive a principled upper bound for the energy matching objective which can be used to fine-tune dLLMs. Experiments on multiple benchmarks in both online and offline setting show that EMBR matches or surpasses the performance of diffu-GRPO and related baselines in the online case, and of DPO in the offline case. Our approach provides a practical alternative for post-training of diffusion LMs.

Energy Matching based Preference Learning for Diffusion Language Models

Large Language Models (LLMs) are increasingly used as judges and reward models in alignment pipelines, where their scores shape learned behavior. Prior work shows these judges can be manipulated by superficial openers (e.g., “Thought process:” or “Let’s solve this step by step.”), but vulnerabilities in intermediate reasoning verification remain underexplored. We identify Fluff Injection, a failure in which a logically necessary step in a chain of reasoning is replaced with plausible-sounding commentary (e.g. “Let’s slow down and check our negatives here”). To measure this failure mode, we introduce FluffInjector, a benchmark of paired minimal examples: for each problem, we generate a GOOD chain and a FLUFF chain that keeps the same step count and final answer while replacing 25-40% of steps with non-inferential filler. Evaluating frontier judges (GPT-4.1, DeepSeek-V3.1, Qwen2.5-7B-Instruct), we find they frequently validate FLUFFED chains, indicating a strong reliance on surface coherence. Using FluffInjector, we fine-tune SmartRM, a verifier trained to emphasize step-to-step logical continuity. SmartRM reduces false positives from 37.43% (GPT-4.1) to 2.68% and achieves 97.27% overall verification accuracy.

FluffInjector: Diagnosing Logical Consistency Failures in Chain-of-Thought Reward Models

Recent advances in Vision–Language Models (VLMs) have demonstrated impressive multimodal understanding in general domains. However, their applicability to decision-oriented domains such as hospitality remains largely unexplored. In this work, we investigate how well VLMs can perform visual question answering (VQA) about hotel and facility images that are central to consumer decision-making. While many existing VQA benchmarks focus on factual correctness, they rarely capture what information users actually find useful. To address this, we first introduce Informativeness as a formal framework to quantify how much hospitality-relevant information an image–question pair provides. Guided by this framework, we construct a new hospitality-specific VQA dataset that covers various facility types, where questions are specifically designed to reflect key user information needs. Using this benchmark, we conduct experiments with several state-of-the-art VLMs, revealing that VLMs are not intrinsically decision-aware-key visual signals remain underutilized, and reliable informativeness reasoning emerges only after modest domain-specific finetuning.

Hospitality-VQA: Decision-Oriented Informativeness Evaluation for Vision–Language Models

The recent rapid real-world adoption of vision-language models (VLMs) raises concerns about how social biases encoded in language may propagate into visual generation. In this work, we examine whether socioeconomic stereotypes, expressed through occupation and income-related linguistic cues in prompts, systematically influences skin-tone representations in text-to-image (T2I) generation, with a focus on colorism as a visual marker of social inequality. We first benchmark 3 small VLMs and 60 human annotators on the Monk Skin Tone (MST) scale using the MST-E dataset. We then conduct a large-scale T2I generation study in which we systematically vary the linguistic framing of income in prompts describing 210 occupations, producing over 2,500 portraits across 3 large VLMs. The skin-tone audit of the portraits by the best-performing annotator (GPT-5 mini) reveals strong color bias: high-income prompts consistently produce lighter-skinned faces, with prompt constraints only modestly attenuating this effect. Bias magnitude varies across generators, with GPT-5 Image-mini and Gemini-2.5 Flash-Image exhibiting more pronounced shifts in MST than Grok-2 Image. Our findings indicate that VLMs encode and amplify ethnoracialized socioeconomic stereotypes in language-conditioned image generation, underscoring the need for cross-modal fairness audits and human-centered evaluations.

Colorism in Multimodal AI: An Empirical Exploration of Socioeconomic Linguistic Bias in Text-to-Image Generation

State Space Models (SSMs) have recently emerged as efficient alternatives to Transformers for sequence modeling, yet extending them to two-dimensional vision tasks remains challenging. The Graph-Generating State Space Model (GG-SSM) addresses this challenge by constructing an adaptive graph, achieving competitive performance on vision benchmarks. However, state propagation over the resulting graph introduces substantial inference overhead, limiting scalability to high-resolution inputs. In this work, we introduce a leaf-guided computation pruning strategy that accelerates GG-SSM inference without modifying the underlying graph topology. Rather than removing nodes or edges, our approach selectively scales or bypasses secondary refinement computations associated with high-dissimilarity leaf nodes, while preserving the low-weight MST backbone. Experiments on multiple long-term time series forecasting benchmarks demonstrate consistent throughput improvements with controlled accuracy degradation across a range of pruning ratios. These results indicate that structure-aware computation pruning is an effective mechanism for improving the scalability of graph-based state space models.

Token Pruning for Improving Graph-Generating State Space Model Performance

Linear recurrent neural networks (LRNNs) provide a structured approach to sequence modeling that bridges classical linear dynamical systems and modern deep learning, offering both expressive power and theoretical guarantees on stability and trainability. In recent years, multiple LRNN-based architectures have been proposed, each introducing distinct parameterizations, discretization schemes, and implementation constraints. However, existing implementations are fragmented across different software frameworks, often rely on framework-specific optimizations, and in some cases require custom CUDA kernels or lack publicly available code altogether. As a result, using, comparing, or extending LRNNs requires substantial implementation effort. To address this, we introduce $\texttt{lrnnx}$, a unified software library that implements several modern LRNN architectures under a common interface. The library exposes multiple levels of control, allowing users to work directly with core components or higher-level model abstractions. $\texttt{lrnnx}$ aims to improve accessibility, reproducibility, and extensibility of LRNN research and applications. We make our code available under a permissive MIT license.

lrnnx: A library for Linear RNNs

This work evaluates the non-English and unstructured text compression performance of Large Language Models (LLMs) by comparing them with traditional baselines on datasets from eight most widely spoken languages. Experimental results show that the evaluated LLM (LLaMA-3.2-1B) was considerably outperformed by the baselines, particularly on non-English datasets, where its performance relative to the best baseline was more than three times worse than on English datasets on average. It also compressed unstructured English data up to more than twofold less effectively than plain English data. Traditional methods, however, remained largely dataset-agnostic. Surprisingly, the LLM achieved worse compression ratios on some datasets than others despite modeling them more accurately. Overall, the outcomes and substantially higher compression time and resource consumption indicate that current LLMs are highly impractical for the compression task, where traditional methods continue to excel. Codes are available at: https://github.com/mehranhaddadi13/llm_compress.

Comparing Text Compression Capabilities of Large Language Models with Traditional Compression Algorithms

Human cognition is deeply intertwined with a sense of time, known as \textit{Chronoception}. This sense allows us to judge how long facts remain valid and when knowledge becomes outdated. Despite progress in vision, language, and motor control, AI still struggles to reason about temporal validity. We introduce Chronocept, the first benchmark to model temporal validity as a continuous probability distribution over time. Using skew-normal curves fitted along semantically decomposed temporal axes, Chronocept captures nuanced patterns of emergence, decay, and peak relevance. It includes two datasets: Benchmark I (atomic facts) and Benchmark II (multi-sentence passages). Annotations show strong inter-annotator agreement (84\% and 89\%). Our baselines predict curve parameters - location, scale, and skewness - enabling interpretable, generalizable learning and outperforming classification-based approaches. Chronocept fills a foundational gap in AI's temporal reasoning, supporting applications in knowledge grounding, fact-checking, retrieval-augmented generation (RAG), and proactive agents. Code and data are publicly available.

Downloads

Next from EACL 2026

How Do Lexical Senses Correspond Between Spoken German and German Sign Language?

Similar lecture

ASSERT: Automated Safety Scenario Red Teaming for Evaluating the Robustness of Large Language Models