Austria

Ensuring the safety alignment of Large Language Models (LLMs) is critical for generating responses consistent with human values. However, LLMs remain vulnerable to jailbreaking attacks, where carefully crafted prompts manipulate them into producing toxic content. One category of such attacks reformulates the task as an optimization problem, aiming to elicit affirmative responses from the LLM. However, these methods heavily rely on predefined objectionable behaviors, limiting their effectiveness and adaptability to diverse harmful queries. In this study, we first identify why the vanilla target loss is suboptimal and then propose enhancements to the loss objective. We introduce DSN (Don’t Say No) attack, which combines a cosine decay schedule method with refusal suppression to achieve higher success rates. Extensive experiments demonstrate that DSN outperforms baseline attacks and achieves state-of-the-art attack success rates (ASR). DSN also shows strong universality and transferability to unseen datasets and black-box models.

ACL 2025

Don&#39;t Say No: Jailbreaking LLM by Suppressing Refusal

learning-based attack

jailbreak

llm safety

### Welcome!
The 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) will take place in Bangkok, Thailand from August 11th to 16th, 2024. Our Virtual Poster Sessions will take place online Thursday, August 22, 2024.

You are required to register for this event. **Please register [here](https://2024.aclweb.org/registration). **

If you have already registered, please check your inbox for an email from Underline granting you access to ACL 2024 content.

Please register!

ACL 2024

The 62nd Annual Meeting of the Association for Computational Linguistics (ACL 2024) will take place in Bangkok, Thailand from August 11th to 16th, 2024. More information will be announced soon.

Don't Say No: Jailbreaking LLM by Suppressing Refusal

poster

### 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**. 

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

As Large Language Models (LLMs) increasingly produce human-quality text, understanding how readers perceive and evaluate AI-generated literature becomes critical. This paper investigates the interplay between measurable textual features and reader preferences in the context of short stories. We analyze multiple datasets where readers (ranging from literary experts to lay audiences) evaluated both human-authored and LLM-generated texts. We propose to model reader's preferences as weighted combinations of intrinsic textual features such as coherence, sentiment volatility, thematic entropy, etc. Results show that (1) when LLM-generated text closely approximates the feature distributions of human text, readers struggle to distinguish them; (2) expert and non-expert readers prioritize distinctly different textual features, revealing systematic variations in literary judgment; and (3) these preference patterns are remarkably stable across diverse datasets. Our framework provides a step towards understanding and potentially tailoring AI-generated literature to meet the expectations of specific audiences.

When AI Writes Like Humans: Capturing the Emergent Patterns of Literary Judgment via Intrinsic Textual Metrics

Large Language Models encode behaviors like refusal within their activation space, but identifying these behaviors remains challenging. Existing methods depend on predefined refusal templates detectable in output tokens or manual review. We introduce **COSMIC** (Cosine Similarity Metrics for Inversion of Concepts), an automated framework for direction selection that optimally identifies steering directions and target layers using cosine similarity, entirely independent of output text. COSMIC achieves steering effectiveness comparable to prior work without any prior knowledge or assumptions of a model’s refusal behavior such as the use of certain refusal tokens. Additionally, COSMIC successfully identifies refusal directions in adversarial scenarios and models with weak safety alignment, demonstrating its robustness across diverse settings.

COSMIC: Generalized Refusal Identification in LLM Activations

We propose IMPARA-GED, a novel reference-free automatic grammatical error correction (GEC) evaluation method with grammatical error detection (GED) capabilities. We focus on the quality estimator of IMPARA, an existing automatic GEC evaluation method, and construct that of IMPARA-GED using a pre-trained language model with enhanced GED capabilities. Experimental results on SEEDA, a meta-evaluation dataset for automatic GEC evaluation methods, demonstrate that IMPARA-GED achieves the highest correlation with human sentence-level evaluations.

IMPARA-GED: Grammatical Error Detection is Boosting Reference-free Grammatical Error Quality Estimator

Psychology research has shown that humans are poor at estimating their performance on tasks, tending towards underconfidence on easy tasks and overconfidence on difficult tasks. We examine three LLMs, Llama-3-70B-instruct, Claude-3-Sonnet, and GPT-4o, on a range of QA tasks of varying difficulty, and show that models exhibit subtle differences from human patterns of overconfidence: less sensitive to task difficulty, and when prompted to answer based on different personas---e.g., expert vs layman, or different race, gender, and ages---the models will respond with stereotypically biased confidence estimations even though their underlying answer accuracy remains the same. Based on these observations, we propose a novel method, Answer-Free Confidence Estimation (AFCE), for improving confidence calibration and LLM interpretability in these settings. AFCE is a self-assessment method that employs two stages of prompting, first eliciting only confidence scores on questions without answers, then asking separately for the answer. Experiments on MMLU and GPQA datasets spanning subjects and difficulty show that AFCE significantly improves confidence calibration performance for the most difficult problems where overconfidence is most pronounced.

Do Language Models Mirror Human Confidence? Exploring Psychological Insights to Address Overconfidence in LLMs

State-of-the-art methods for grammatical error correction require large amounts of parallel data for training. Due to lack of gold-labeled data, techniques that create synthetic training data have become popular. We show that models trained on such data tend to correct a limited range of grammar and spelling mistakes that involve character-level changes, but perform poorly on (more complex) phenomena that require word-level changes. We propose to address the gap in performance on such errors by generating synthetic data through selective data augmentation via round-trip machine translation. We show that the proposed technique, SeLex-RT is capable of generating mistakes that are similar to those observed with language learners. Using the approach with two types of state-of-the-art learning frameworks and two languages (Russian and Ukrainian), we achieve substantial improvements, compared to training on synthetic data produced with standard techniques. The method is especially effective in low-resource settings. Analysis of the output reveals that models trained on data noisified using round-trip translation are capable of making word-level changes and correct lexical errors common with language learners.

Low-Resource Grammatical Error Correction: Selective Data Augmentation with Round-Trip Machine Translation

Authorship verification (AV) is a crucial task for applications like identity verification, plagiarism detection, and AI-generated text identification. However, datasets for training and evaluating AV models are primarily in English and primarily in a single domain. This precludes analysis of AV techniques for generalizability and can cause seemingly valid AV solutions to, in fact, rely on topic-based features rather than actual authorship features. To address this limitation, we introduce the Million Authors Corpus (), a novel dataset encompassing contributions from dozens of languages on Wikipedia. It includes only long and contiguous textual chunks taken from Wikipedia edits and links those texts to their authors. includes 60.08M textual chunks, contributed by 1.29M Wikipedia authors. It enables broad-scale cross-lingual and cross-domain AV evaluation to ensure accurate analysis of model capabilities that are not overly optimistic. We provide baseline evaluations using state-of-the-art AV models as well as information retrieval models that are not AV-specific in order to demonstrate 's unique cross-lingual and cross-domain ablation capabilities.

The Million Authors Corpus: A Cross-Lingual and Cross-Domain Wikipedia Dataset for Authorship Verification

Ensuring persona fidelity in large language models (LLMs) is essential for maintaining coherent and engaging human-AI interactions. However, LLMs often exhibit Out-of-Character (OOC) behavior, where generated responses deviate from an assigned persona, leading to inconsistencies that affect model reliability. Existing evaluation methods typically assign single scores to entire responses, struggling to capture subtle persona misalignment, particularly in long-form text generation. To address this limitation, we propose an atomic-level evaluation framework that quantifies persona fidelity at a finer granularity. Our three key metrics measure the degree of persona alignment and consistency within and across generations. Our approach enables a more precise and realistic assessment of persona fidelity by identifying subtle deviations that real users would encounter. Through our experiments, we demonstrate that our framework effectively detects persona inconsistencies that prior methods overlook. By analyzing persona fidelity across diverse tasks and personality types, we reveal how task structure and persona desirability influence model adaptability, highlighting challenges in maintaining consistent persona expression.

Spotting Out-of-Character Behavior: Atomic-Level Evaluation of Persona Fidelity in Open-Ended Generation

Using LLMs for Multi-Document Topic Extraction has recently gained popularity due to their apparent high-quality outputs, expressiveness, and ease of use. However, most existing evaluation practices are not designed for LLM-generated topics and result in low inter-annotator agreement scores, hindering the reliable use of LLMs for the task. To address this, we introduce T5Score, an evaluation methodology that decomposes the quality of a topic set into quantifiable aspects, measurable through easy-to-perform annotation tasks. This framing enables a convenient, manual or automatic, evaluation procedure resulting in a strong inter-annotator agreement score.To substantiate our methodology and claims, we perform extensive experimentation on multiple datasets and report the results.

T5Score: A Methodology for Automatically Assessing the Quality of LLM Generated Multi-Document Topic Sets

Large Language Models (LLMs) exhibit remarkable multilingual generalization despite being predominantly trained on English-centric corpora. A fundamental question arises: how do LLMs achieve such robust multilingual capabilities? We take the case of non-Roman script languages, we investigate the role of Romanization—the representation of non-Roman scripts using Roman characters—as a bridge in multilingual processing. Using mechanistic interpretability techniques, we analyze next-token generation and find that intermediate layers frequently represent target words in Romanized form before transitioning to native script, a phenomenon we term Latent Romanization. Further, through activation patching experiments, we demonstrate that LLMs encode semantic concepts similarly across native and Romanized scripts, suggesting a shared underlying representation. Additionally, for translation into non-Roman script languages, our findings reveal that when the target language is in Romanized form, its representations emerge earlier in the model’s layers compared to native script. These insights contribute to a deeper understanding of multilingual representation in LLMs and highlight the implicit role of Romanization in facilitating language transfer.

RomanLens: The Role Of Latent Romanization In Multilinguality In LLMs

We introduce LLM-as-an-Interviewer, a novel paradigm for evaluating large language models (LLMs). This approach leverages multi-turn interactions where the LLM interviewer actively provides feedback on responses and poses follow-up questions to the evaluated LLM. At the start of the interview, the LLM interviewer dynamically modifies datasets to generate initial questions, mitigating data contamination. We apply the LLM-as-an-Interviewer framework to evaluate six models on the reasoning, factuality and instruction-following tasks. Our results show that the framework effectively provides insights into LLM performance, including the quality of initial responses, adaptability to feedback, and ability to address follow-up queries like clarification or additional knowledge requests. The framework also addresses key limitations of conventional methods like LLM-as-a-Judge, including verbosity bias and inconsistency across runs. Finally, we propose the Interview Report, which aggregates insights from the interview process, providing examples and a comprehensive analysis of the LLM’s strengths and weaknesses. This report offers a detailed snapshot of the model's real-world applicability.

Downloads

Next from ACL 2025

When AI Writes Like Humans: Capturing the Emergent Patterns of Literary Judgment via Intrinsic Textual Metrics

Similar lecture

SpeechGuard: Exploring the Adversarial Robustness of Multi-modal Large Language Models

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES