Morocco

Large Language Models (LLMs) have achieved substantial progress in alignment, ensuring safer and more reliable outputs. However, jailbreak attacks can still bypass these safeguards and provoke harmful responses from well-aligned models. While some studies have achieved defenses against jailbreak attacks by modifying output distributions or detecting harmful content, the exact rationale still remains elusive. In this work, we present a novel neuron-level interpretability method that focuses on the role of safety-related knowledge neurons. Unlike existing approaches, our method projects the model&#39;s internal representation into a more consistent and interpretable vocabulary space. We then show that adjusting the activation of safety-related neurons can effectively control the model&#39;s behavior with a mean ASR higher than 97%. Building on this insight, we propose SafeTuning, a fine-tuning strategy that reinforces safety-critical neurons to improve model robustness against jailbreaks. SafeTuning consistently reduces attack success rates across multiple LLMs and outperforms all four baseline defenses. These findings offer a new perspective on understanding and defending against jailbreak attacks.

EACL 2026

Unraveling LLM Jailbreaks Through Safety Knowledge Neurons

jailbreak attacks

model interpretability

large language models

### Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

Message from the General Chair: 
*It is my great pleasure and honor to welcome you to the 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025), held in beautiful Vienna, Austria, from July 27 to August 1, 2025. ACL2025continues our field’s tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.* 
[Read more](https://drive.google.com/file/d/1GI_hvOpjswAuYdUTromfeDiPpCcqidwg/view?usp=sharing)

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 ACL 2025". 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 need to log in with the **email address you registered with**. 

ACL 2025

Welcome to The 63rd Annual Meeting of the Association for Computational Linguistics (ACL 2025)

Large Language Models (LLMs) have achieved substantial progress in alignment, ensuring safer and more reliable outputs. However, jailbreak attacks can still bypass these safeguards and provoke harmful responses from well-aligned models. While some studies have achieved defenses against jailbreak attacks by modifying output distributions or detecting harmful content, the exact rationale still remains elusive. In this work, we present a novel neuron-level interpretability method that focuses on the role of safety-related knowledge neurons. Unlike existing approaches, our method projects the model's internal representation into a more consistent and interpretable vocabulary space. We then show that adjusting the activation of safety-related neurons can effectively control the model's behavior with a mean ASR higher than 97%. Building on this insight, we propose SafeTuning, a fine-tuning strategy that reinforces safety-critical neurons to improve model robustness against jailbreaks. SafeTuning consistently reduces attack success rates across multiple LLMs and outperforms all four baseline defenses. These findings offer a new perspective on understanding and defending against jailbreak attacks.

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.

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. 

The advent of Large Language Models (LLMs) has significantly advanced web-based Question Answering (QA) systems over semi-structured content, raising questions about the continued utility of knowledge extraction for question answering. This paper investigates the value of triple extraction in this new paradigm by extending an existing benchmark with knowledge extraction annotations and evaluating commercial and open-source LLMs of varying sizes. Our results show that web-scale knowledge extraction remains a challenging task for LLMs. Despite achieving high QA accuracy, LLMs can still benefit from knowledge extraction, through augmentation with extracted triples and multi-task learning. These findings provide insights into the evolving role of knowledge triple extraction in web-based QA and highlight strategies for maximizing LLM effectiveness across different model sizes and resource settings.

Knowledge Extraction on Semi-Structured Content: Does It Remain Relevant for Question Answering in the Era of LLMs?

Visual metonymy is a cognitive mechanism through which a concept is evoked indirectly by presenting associated visual cues rather than explicitly depicting it. These associative links prompt viewers to infer broader concepts. In this work, we present the first computational investigation of visual metonymy. We introduce a novel pipeline grounded in semiotic theory, that leverages large language models and text-to-image models to generate metonymic visual representations. Using this framework, we construct ViMET, the first visual metonymy dataset comprising 2,000 multiple-choice questions to evaluate the cognitive reasoning abilities in multimodal language models. Experimental results on our dataset reveal a significant gap between human performance (86.9%) and state-of-the-art vision-language models (65.9%), highlighting limitations in machines ability to interpret indirect visual references.

Inferring the Unseen: A Computational Approach to Visual Metonymy

Large language model (LLM)-based systems now power chatbots, computer-use agents, and even robots, relying on prompts that often require costly manual tuning. LLM-based prompt optimizers cut that effort by iteratively refining prompts with scored feedback, yet their security has gone largely unexamined. Our work provides the first systematic analysis of poisoning risks in this optimization loop, revealing that optimization itself can become a new vulnerability point. Using HarmBench, we show optimized systems are far more vulnerable to manipulated feedback than to malicious queries, with Attack Success Ratio jumping by up by 48% absolute points. We propose a feedback-poisoning attack that embeds fake reward tokens in prompts, significantly boosting attack success under realistic assumptions, and evaluate tailored defenses that drastically decrease that threat. Our findings establish prompt optimization pipelines as an urgent and underexplored attack surface for LLM safety research, calling for stronger safeguards in future optimization frameworks.

Are My Optimized Prompts Compromised? Exploring Vulnerabilities of LLM-based Optimizers

Health wellness agents typically rely on large language models (LLMs) for response generation, where contextual information from a user's conversation history can be used for response grounding and personalization. High-quality conversation summaries are one such method which can reduce the number of input tokens during response generation, decreasing overhead and inference latency. However, directly purposing LLMs for this task is infeasible due to the scale of the task, the compute overhead, and health data compliance regulations. Furthermore, ground truth for real-world datasets is scarce due to privacy concerns and the high cost of health expert annotators. These factors necessitate the development of small, potentially on-device, language models capable of health dialogue summarization, particularly in the absence of ground truth labels. In this paper, we first present a comprehensive empirical study that benchmarks a variety of state-of-the-art smaller language models to better understand their baseline capabilities. Second, we present an unsupervised method that uses the summaries from multiple models, refined with iterative feedback, to generate high-quality summaries of health dialogues. Experiments illustrate that our method is able to outperform baseline on both open-source and proprietary benchmarks. Notably, our method can be run viably on local compute without a GPU, using just a single Macbook with 16 GB of memory.

Compact Language Models with Iterative Text Refinement for Health Dialogue Summarization

There are many methods and systems to tackle the ontology alignment problem, yet a major challenge persists in producing high-quality mappings among a set of input ontologies. Adopting a human-in-the-loop approach during the alignment process has become essential in applications requiring very accurate mappings. However, user involvement is expensive when dealing with large ontologies. In this paper, we evaluate the feasibility of using Large Language Models (LLM) to aid the ontology alignment problem. The use of LLMs is focused only on the validation of a subset of correspondences where an ontology alignment system (e.g., LogMap) is very uncertain. We have conducted an extensive analysis over several tasks of the Ontology Alignment Evaluation Initiative (OAEI), reporting in this paper the performance of several state-of-the-art LLMs using different ontology-driven prompt templates. In the OAEI 2025 Bio-ML track, LogMap with an LLM-based Oracle has achieved the top-2 overall results. LLM efficacy is also assessed against simulated Oracles with varying error rates.

Large Language Models as Oracles for Ontology Alignment

Most realistic task automation problems require large language models (LLMs) to call tools, which often return complex JSON responses. These responses must be further processed to derive the information necessary for task completion. The ability of LLMs to do so is under-studied. In this paper, we study the tool response processing task and LLMs' abilities to process structured (JSON) responses. We created a dataset for this task, and evaluated 15 open and closed weight models using multiple prompting approaches. Our results show that JSON processing remains a difficult task even for frontier models across multiple prompting strategies. The optimal response processing strategy depends on both the nature and size of the tool outputs, as well as the complexity of the required reasoning. Variations in processing approaches can lead to performance differences ranging from 3% to 50%.

How Good Are LLMs at Processing Tool Outputs?

The factual reliability of Large Language Models (LLMs) remains a critical barrier to their adoption in high-stakes domains due to their propensity to hallucinate. Current detection methods often rely on surface-level signals from the model's output, overlooking the failures that occur within the model's internal reasoning process. In this paper, we introduce a new paradigm for hallucination detection by analyzing the dynamic topology of the evolution of model's layer-wise attention. We model the sequence of attention matrices as a zigzag graph filtration and use zigzag persistence, a tool from Topological Data Analysis, to extract a topological signature. Our core hypothesis is that factual and hallucinated generations exhibit distinct topological signatures. We validate our framework, HalluZig, on multiple benchmarks, demonstrating that it outperforms strong baselines. Furthermore, our analysis reveals that these topological signatures are generalizable across different models and can enable early detection of hallucinations.

HalluZig: Hallucination Detection using Zigzag Persistence

Large Language Models and Vision-Language Models have achieved impressive performance across a wide range of tasks, yet they remain vulnerable to carefully crafted perturbations. In this study, we seek to pinpoint the sources of this fragility by identifying parameters and input dimensions (pixels or token embeddings) that are susceptible to such perturbations. To this end, we propose a stability measure called FI, First order local Influence, which is rooted in information geometry and quantifies the sensitivity of individual parameter and input dimensions. Our extensive analysis across LLMs and VLMs (from 1.5B to 13B parameters) reveals that: (I) A small subset of parameters or input dimensions with high FI values disproportionately contribute to model brittleness. (II) Mitigating the influence of these vulnerable parameters during model merging leads to improved performance.

Breach in the Shield: Unveiling the Vulnerabilities of Large Language Models

We introduce Argument Representation Coverage (ARC), a bottom-up evaluation framework that assesses how well summaries preserve structured salient arguments, a crucial issue in summarizing high-stakes domains such as law. ARC provides an interpretable lens by distinguishing between different information types to be covered and by separating omissions from factual errors.Using ARC, we evaluate summaries from eight open-weight LLMs in two domains where argument roles are central: long legal opinions and scientific articles. Our results show that while LLMs capture some salient roles, they frequently omit critical information, particularly when arguments are sparsely distributed across the input. Moreover, ARC uncovers systematic patterns—showing how context window positional bias and role-specific preferences shape argument coverage—providing actionable guidance for developing more complete and reliable summarization strategies.

ARC: Argument Representation and Coverage Analysis for Zero-Shot Long Document Summarization with Instruction Following LLMs

Current speech evaluation suffers from two critical limitations: the need and difficulty of designing specialized systems targeting individual audio characteristics, and poor correlation between automatic evaluation methods and human preferences. This work presents a systematic study of Large Audio Model (LAM) as a Judge, AudioJudge, investigating whether it can provide a unified evaluation framework that addresses both challenges. We systematically explore AudioJudge across audio characteristic detection tasks, including pronunciation, speaking rate, speaker identification and speech quality, and system-level human preference simulation for automated benchmarking. We investigate different prompt engineering strategies, finding that audio concatenation combined with in-context learning significantly improves performance across both audio characteristic detection and human preference simulation tasks. We further introduce a multi-aspect ensemble AudioJudge to enable general-purpose multi-aspect audio evaluation. This method decomposes speech assessment into specialized judges for lexical content, speech quality, and paralinguistic features, achieving up to 0.91 Spearman correlation with human preferences on our system ranking benchmark. Robustness analysis reveals that while LAMs maintain strong performance under acoustic noise, they exhibit significant verbosity and positional biases that require careful mitigation.

Downloads

Next from EACL 2026

Knowledge Extraction on Semi-Structured Content: Does It Remain Relevant for Question Answering in the Era of LLMs?

Similar lecture

JailbreakRadar: Comprehensive Assessment of Jailbreak Attacks Against LLMs

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from EACL 2026

Knowledge Extraction on Semi-Structured Content: Does It Remain Relevant for Question Answering in the Era of LLMs?

Similar lecture

JailbreakRadar: Comprehensive Assessment of Jailbreak Attacks Against LLMs

Downloads