Alignment Quality Index (AQI) : Beyond Refusals: AQI as an Intrinsic Alignment Diagnostic via Latent Geometry, Cluster Divergence, and Layer wise Pooled Representations (2025.emnlp-main)
Copied to clipboard
Abhilekh Borah, Chhavi Sharma, Danush Khanna, Utkarsh Bhatt, Gurpreet Singh, Hasnat Md Abdullah, Raghav Kaushik Ravi, Vinija Jain, Jyoti Patel, Shubham Singh, Vasu Sharma, Arpita Vats, Rahul Raja, Aman Chadha, Amitava Das
| Challenge: | a new metric measures the quality of large language models (LLMs) that detects hidden misalignments and jailbreak risks. |
| Approach: | They propose a decoding-invariant metric that measures latent safety failures . they propose 'Alignment Quality Index' to measure latent activations in latent space . |
| Outcome: | The proposed metric detects latent safety failures overlooked by behavioral benchmarks and jailbreaks. |
Similar Papers
The Unintended Trade-off of AI Alignment: Balancing Hallucination Mitigation and Safety in LLMs (2026.findings-eacl)
Copied to clipboard
| Challenge: | Hallucination in large language models has been studied, but a side effect remains unrecognized . a new study examines the trade-off between truthfulness and safety alignment . |
| Approach: | They propose a method that disentangles hallucination from hallucinian features using sparse autoencoders. |
| Outcome: | The proposed method preserves refusal behavior and task utility while maintaining safety alignment. |
How Alignment and Jailbreak Work: Explain LLM Safety through Intermediate Hidden States (2024.findings-emnlp)
Copied to clipboard
| Challenge: | Large language models (LLMs) rely on safety alignment to avoid malicious user inputs. |
| Approach: | They employ weak classifiers to explain LLM safety through the intermediate hidden states. |
| Outcome: | The proposed model can identify malicious and normal inputs and detect malicious ones without jailbreak. |
Fake Alignment: Are LLMs Really Aligned Well? (2024.naacl-long)
Copied to clipboard
Yixu Wang, Yan Teng, Kexin Huang, Chengqi Lyu, Songyang Zhang, Wenwei Zhang, Xingjun Ma, Yu-Gang Jiang, Yu Qiao, Yingchun Wang
| Challenge: | Existing studies on large language models have shown that they are poorly aligned in practice. |
| Approach: | They propose a framework to evaluate safety in large language models . they propose two new metrics to quantify fake alignment and obtain corrected performance estimation. |
| Outcome: | The proposed framework and two metrics show that some models with purported safety are poorly aligned in practice. |
Probing the Safety Robustness of LLMs in Latent Space (2026.acl-long)
Copied to clipboard
Tianle Gu, Kexin Huang, Zongqi Wang, Yixu Wang, Jie Li, Xin Wang, Yang Yao, Yujiu Yang, Yan Teng, Yingchun Wang
| Challenge: | Despite substantial progress in safety alignment techniques, aligned large language models can still produce unsafe responses under minor internal perturbations. |
| Approach: | They introduce Activation Steering Attack (ASA) and leverage the Negative Log-Likelihood (NLL) as a diagnostic signal to probe the local sensitivity of safety behaviors in latent space. |
| Outcome: | The proposed method is model-agnostic and supervision-free, enabling a general and reproducible diagnostic metric for analyzing safety robustness. |
The Art of (Mis)alignment: How Fine-Tuning Methods Effectively Misalign and Realign LLMs in Post-Training (2026.findings-acl)
Copied to clipboard
Rui Zhang, Hongwei Li, Yun Shen, Xinyue Shen, Wenbo Jiang, Guowen Xu, Yang Liu, Michael Backes, Yang Zhang
| Challenge: | Misaligned large language models can magnify harm by exploiting them to undermine safety . et al., 2022b; Bai e.t., 2023): misalignment, realignment and model-specific resistance are important . |
| Approach: | They evaluate four methods to identify a mechanism asymmetry between attack and defense . they find that ORPO is most effective for misalignment, but DPO excels in realignment . |
| Outcome: | The proposed methods show a mechanism asymmetry between attack and defense . the proposed methods excel in realignment, but at the expense of model utility . |
Language Models Resist Alignment: Evidence From Data Compression (2025.acl-long)
Copied to clipboard
Jiaming Ji, Kaile Wang, Tianyi Alex Qiu, Boyuan Chen, Jiayi Zhou, Changye Li, Hantao Lou, Josef Dai, Yunhuai Liu, Yaodong Yang
| Challenge: | Large language models (LLMs) may exhibit undesirable behaviors due to the inevitable biases and harmful content present in training. |
| Approach: | They propose to investigate the elasticity of large language models by examining their performance. |
| Outcome: | The proposed model performance declines rapidly before reverting to the pre-training distribution, the authors show . the proposed model weight and code are available at pku-lm-res ist-alignment.github.io. |
AlignBench: Benchmarking Chinese Alignment of Large Language Models (2024.acl-long)
Copied to clipboard
Xiao Liu, Xuanyu Lei, Shengyuan Wang, Yue Huang, Andrew Feng, Bosi Wen, Jiale Cheng, Pei Ke, Yifan Xu, Weng Lam Tam, Xiaohan Zhang, Lichao Sun, Xiaotao Gu, Hongning Wang, Jing Zhang, Minlie Huang, Yuxiao Dong, Jie Tang
| Challenge: | Effective evaluation of alignment for emerging Chinese LLMs is still significantly lacking, calling for real-scenario grounded, open-ended, challenging and automatic evaluations tailored for alignment. |
| Approach: | They propose a multi-dimensional benchmark for evaluating LLMs’ alignment in Chinese with 8 main categories, 683 real-scenario rooted queries and corresponding human verified references. |
| Outcome: | The benchmark uses a human-in-the-loop data curation pipeline, 683 real-scenario rooted queries and human verified references. |
Into the Gray Zone: Domain Contexts Can Blur LLM Safety Boundaries (2026.acl-long)
Copied to clipboard
Ki Sen Hung, Xi Yang, Chang Liu, Haoran Li, Kejiang Chen, Changxuan Fan, Tsun On Kwok, Weiming Zhang, Xiaomeng Li, Yangqiu Song
| Challenge: | a goal of LLM alignment is to balance usefulness with harmlessness, but this conflictes when knowledge serves both legitimate and malicious purposes. |
| Approach: | They propose a framework that combines safety-research contexts with adversarial interactions to exploit a vulnerability in Jargon queries. |
| Outcome: | a framework outperforms existing methods in analyzing Jargon queries, a study shows . it achieves 93% of attacks across seven models, while remaining useful, the authors say . |
When to Trust LLMs: Aligning Confidence with Response Quality (2024.findings-acl)
Copied to clipboard
Shuchang Tao, Liuyi Yao, Hanxing Ding, Yuexiang Xie, Qi Cao, Fei Sun, Jinyang Gao, Huawei Shen, Bolin Ding
| Challenge: | Existing methods express reliability by confidence level, but lack objective guidance . Existing approaches express reliability but lack guidance on when to trust LLMs . |
| Approach: | They propose a reward-based approach to align confidence with quality to ensure reliability . they propose 'conqORD' to help model to verbalize greater confidence for higher quality responses . |
| Outcome: | Experiments show that CONQORD significantly improves confidence and response accuracy . the proposed approach can be used to determine reliability of large language models . |
Health-ORSC-Bench: A Benchmark for Measuring Over-Refusal and Safety Completion in Health Context (2026.findings-acl)
Copied to clipboard
| Challenge: | Existing safety alignment benchmarks fail to evaluate Safe Completion: the model’s ability to maximise helpfulness on dual-use or borderline queries without crossing into actionable harm. |
| Approach: | They propose a large-scale benchmark to measure Over-Refusal and Safe Completion quality in healthcare. |
| Outcome: | The framework evaluates 30 state-of-the-art LLMs including GPT-5 and Claude-4. |