Refusal in Language Models Is Mediated by a Single Direction
The DOI matches its Crossref record; 18 of the 18 references it cites were found in a registry or an index.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“ref. 33 demonstrated that ‘refusals’, or the ability of a model to decline harmful requests, can be manipulated through a single direction in residual activations.”
From Training large language models on narrow tasks can lead to broad misalignment (2026) · cites this paper as [33]
In the citing paper
“Importantly, this suggests that harmlessness training (which trains models to refuse to answer harmful requests) may be cheaply undone.”
From Open Problems in Mechanistic Interpretability (2025) · cites this paper as Arditi et al., 2024
In the citing paper
“By understanding the mechanisms of jailbreaking, we can anticipate the means through which a user might bypass existing safeguards (Lee et al., 2025;Arditi et al., 2024).”
From Open Problems in Mechanistic Interpretability (2025) · cites this paper as Arditi et al., 2024
Citations
Cites
The publisher hasn’t deposited this paper’s reference list.
Cited by
Items 201–300 of 1,000
Applying Refusal-Vector Ablation to Llama 3.1 70B AgentsPreprint
Lermen, S., Dziemian, M., & Pimpale, G. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2410.10871 · Cited by 13
LoRA is All You Need for Safety Alignment of Reasoning LLMsPreprint
Xue, Y., & Mirzasoleiman, B. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2507.17075 · Cited by 13
RepIt: Steering Language Models with Concept-Specific Refusal Vectors
Siu, V., Henry, N. W., Crispino, N., Liu, Y., Song, D. X., & Wang, C. (2026) · International Conference on Learning Representations · doi:10.48550/arxiv.2509.13281 · Cited by 13
Towards Understanding and Improving Refusal in Compressed Models via Mechanistic InterpretabilityPreprint
Chhabra, V. K., & Khalili, M. M. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2504.04215 · Cited by 14
Ding, C., Wu, J., Sheng, L., Zhang, F., Yuan, Y., Wang, X., & He, X. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2510.04217 · Cited by 13
Comparative Analysis of LLM Abliteration Methods: A Cross-Architecture Evaluation
Young, R. J. (2025) · arXiv
The Geometry of Self-Verification in a Task-Specific Reasoning ModelPreprint
Lee, A., Sun, L., Wendler, C., Viégas, F. B., & Wattenberg, M. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2504.14379 · Cited by 13
Activation Steering for Masked Diffusion Language ModelsPreprint
Shnaidman, A., Feiglin, E., Yaari, O., Mentel, E., LeVi, A., & Lapid, R. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2512.24143 · Cited by 13
Closing the Confidence-Faithfulness Gap in Large Language ModelsPreprint
Miao, M. M., & Ungar, L. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.25052 · Cited by 13
Estimating Tail Risks in Language Model Output DistributionsPreprint
Angell, R., Singhal, R., Horvitz, Z., Yu, Z., Ranganath, R., McKeown, K. R., & He, H. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.22167 · Cited by 13
AGD: Adversarial Game Defense Against Jailbreak Attacks in Large Language Models
Pan, S., Tian, Z., Huang, Z., Yu, W., Wen, Z., Liu, X., Lü, K., Huang, M., & Li, D. (2025) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2025.acl-long.851 · Cited by 13
Uncensored AI in the Wild: Tracking Publicly Available and Locally Deployable LLMs
Sokhansanj, B. A. (2025) · Future Internet · doi:10.3390/fi17100477 · Cited by 13
From Weights to Activations: Is Steering the Next Frontier of Adaptation?
Ostermann, S., Gurgurov, D., Baeumel, T., Hedderich, M. A., Lapuschkin, S., Samek, W., & Schmitt, V. (2026) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2026.acl-long.1377 · Cited by 12
Probing the Difficulty Perception Mechanism of Large Language ModelsPreprint
Lee, S., Yin, Q., Leong, C. T., Zhang, J., Gong, Y., Ni, S., Yang, M., & Shen, X. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2510.05969 · Cited by 12
\texttt{R$^\textbf{2}$AI}: Towards Resistant and Resilient AI in an Evolving WorldPreprint
Sun, Y., Wang, X., Fu, J., Lu, C., & Zhou, B. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2509.06786 · Cited by 12
Re-Emergent Misalignment: How Narrow Fine-Tuning Erodes Safety Alignment in LLMsPreprint
Giordani, J. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2507.03662 · Cited by 12
Linear Representation Transferability Hypothesis: Leveraging Small Models to Steer Large ModelsPreprint
Bello, F., Das, A., Zeng, F., Yin, F., & Leqi, L. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2506.00653 · Cited by 12
Investigating CoT Monitorability in Large Reasoning ModelsPreprint
Yang, S., Wu, J., Gong, X., Wu, X., Wong, D., Liu, N., & Wang, D. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2511.08525 · Cited by 12
Towards Safeguarding LLM Fine-tuning APIs against Cipher AttacksPreprint
Youstra, J., Mahfoud, M., Yan, Y., Sleight, H., Perez, E., & Sharma, M. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2508.17158 · Cited by 12
Latent Structure of Affective Representations in Large Language ModelsPreprint
Choi, B., & Weber, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.07382 · Cited by 12
Your Agentic LLMs Secretly Encode Indirect Prompt-Injection Exposure in Hidden StatesPreprint
Dong, J., Liu, Y., Zhang, M., Deng, N., Peng, X., Zhang, X., Zhang, T., Zhang, J., & Qiu, H. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2608.02657 · Cited by 12
The Alignment Veto: How Safety Training Suppresses Cultural Knowledge in LLMs
Zahraei, P. S., Tur, G., Hakkani-Tür, D., & Asgari, E. (2026) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/2026.emnlp-main.1874 · Cited by 11
Do We Know What LLMs Don’t Know? A Study of Consistency in Knowledge Probing
Zhao, R., Köksal, A., Modarressi, A., Hedderich, M. A., & Schuetze, H. (2025) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2025.findings-emnlp.1263 · Cited by 11
How Does DPO Reduce Toxicity? A Mechanistic Neuron-Level Analysis
Yang, Y., Sondej, F., Mayne, H., Lee, A., & Mahdi, A. (2025) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2025.emnlp-main.1501 · Cited by 11
Unsupervised Concept Vector Extraction for Bias Control in LLMs
Cyberey, H., Ji, Y., & Evans, D. (2025) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2025.emnlp-main.1439 · Cited by 11
Refusal Steering: Fine-grained Control over LLM Refusal Behaviour for Sensitive Topics
García-Ferrero, I., Montero, D., & Orús, R. (2026) · Language Resources and Evaluation Conference (LREC) · doi:10.63317/2zypcp2r538o · Cited by 11
Liu, X., Wang, X., Zhang, Y., Kariyappa, S., Xiang, C., Chen, M., Suh, G. E., & Xiao, C. (2026) · ACM SIGSAC Conference on Computer and Communications Security (CCS) · doi:10.48550/arxiv.2602.00154 · Cited by 12
Goyal, N., & Daumé, H. (2026) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2026.eacl-long.268 · Cited by 11
SafeConstellations: Mitigating Over-Refusals in LLMs Through Task-Aware Representation Steering
Maskey, U., Yadav, S., Dras, M., & Naseem, U. (2026) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2026.acl-long.2056 · Cited by 11
Why do LLaVA Vision-Language Models Reply to Images in English?Preprint
Hinck, M., Holtermann, C., Olson, M., Schneider, F., Yu, S., Bhiwandiwalla, A., Lauscher, A., Tseng, S.-Y., & Lal, V. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2407.02333 · Cited by 11
Learning Safety Constraints for Large Language ModelsPreprint
Chen, X., As, Y., & Krause, A. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2505.24445 · Cited by 11
TwinBreak: Jailbreaking LLM Security Alignments based on Twin PromptsPreprint
Krauß, T., Dashtbani, H., & Dmitrienko, A. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2506.07596 · Cited by 11
Video Unlearning via Low-Rank Refusal Vector
Facchiano, S., Saravalle, S., Migliarini, M., Matteis, E. D., Sampieri, A., Pilzer, A., Rodolà, E., Spinelli, I., Franco, L., & Galasso, F. (2025) · International Conference on Learning Representations · doi:10.48550/arxiv.2506.07891 · Cited by 11
Internal Causal Mechanisms Robustly Predict Language Model Out-of-Distribution BehaviorsPreprint
Huang, J., Tao, J., Icard, T., Yang, D., & Potts, C. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2505.11770 · Cited by 11
Beyond I'm Sorry, I Can't: Dissecting Large Language Model RefusalPreprint
Prakash, N., Jie, Y. W., Abdullah, A., Satapathy, R., Cambria, E., & Lee, R. K.-W. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2509.09708 · Cited by 12
Linear Personality Probing and Steering in LLMs: A Big Five StudyPreprint
Frising, M., & Balcells, D. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2512.17639 · Cited by 11
Old Habits Die Hard: How Conversational History Geometrically Traps LLMsPreprint
Simhi, A., Barez, F., Tutek, M., Belinkov, Y., & Cohen, S. B. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.03308 · Cited by 11
Local Linearity of LLMs Enables Activation Steering via Model-Based Linear Optimal ControlPreprint
Skifstad, J., Yang, X. A., & Chou, G. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.19018 · Cited by 12
How's it going? Reinforcement learning in language models recruits a functional welfare axisPreprint
Han, A. Q., Chalmers, D. J., & Izmailov, P. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.30232 · Cited by 11
Improving Instruction-Following in Language Models through Activation Steering
Stolfo, A., Balachandran, V., Yousefi, S., Horvitz, E., & Nushi, B. (2024) · arXiv · doi:10.48550/arxiv.2410.12877 · Cited by 10
IF-Guide: Influence Function-Guided Detoxification of LLMs
Coalson, Z., Bae, J., Carlini, N., & Hong, S. (2025) · neural information processing systems · doi:10.52202/085713-2638 · Cited by 11
Oldfield, J., Im, S., Li, S., Nicolaou, M. A., Patras, I., & Chrysos, G. G. (2025) · neural information processing systems · doi:10.52202/085713-4836 · Cited by 10
Wu, J., Chen, C., Hou, C., & Yuan, X. (2025) · Findings of the Association for Computational Linguistics: EMNLP (Findings of EMNLP) · doi:10.18653/v1/2025.findings-emnlp.450 · Cited by 10
A Simple Yet Effective Method for Non-Refusing Context Relevant Fine-grained Safety Steering in LLMs
Ghosh, S., Bhattacharjee, A., Ziser, Y., & Parisien, C. (2025) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/v1/2025.emnlp-main.1781 · Cited by 10
In-Context Representation Hijacking
Yona, I., Sarid, A., Karasik, M., & Gandelsman, Y. (2026) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2026.acl-long.768 · Cited by 11
Bleeding Pathways: Vanishing Discriminability in LLM Hidden States Fuels Jailbreak Attacks
Zhang, Y., Liu, T., Zhao, Z., Meng, G., & Chen, K. (2026) · Network and Distributed System Security Symposium · doi:10.14722/ndss.2026.240004 · Cited by 10
Improving Multilingual Language Models by Aligning Representations through Steering
Mahmoud, O., Semage, B. L., Karimpanal, T. G., & Rana, S. (2026) · Language Resources and Evaluation Conference (LREC) · doi:10.63317/244dsoue8zu2 · Cited by 10
Leong, C. T., Yin, Q., Wang, J., & Li, W. (2025) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2025.acl-long.738 · Cited by 10
Model-Editing-Based Jailbreak against Safety-aligned Large Language ModelsPreprint
Li, Y., Zhang, Z., Wang, K., Shi, L., & Wang, H. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2412.08201 · Cited by 10
Concept-ROT: Poisoning Concepts in Large Language Models with Model Editing
Grimes, K., Christiani, M., Shriver, D., & Connor, M. (2024) · International Conference on Learning Representations · doi:10.48550/arxiv.2412.13341 · Cited by 10
Differentially Private Steering for Large Language Model Alignment
Goel, A., Hu, Y., Gurevych, I., & Sanyal, A. (2025) · International Conference on Learning Representations · doi:10.48550/arxiv.2501.18532 · Cited by 10
Disentangled Safety Adapters Enable Efficient Guardrails and Flexible Inference-Time AlignmentPreprint
Krishna, K. S., Cheng, J. Y., Maalouf, C., & Gatys, L. A. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2506.00166 · Cited by 10
Anchoring Refusal Direction: Mitigating Safety Risks in Tuning via Projection ConstraintPreprint
Du, Y., Fan, F., Zhao, S., Cao, J., Lin, Q., He, K., Liu, T., Qin, B., & Feng, M. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2509.06795 · Cited by 10
Effective and Efficient Jailbreaks of Black-Box LLMs with Cross-Behavior AttacksPreprint
Gohil, V. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2503.08990 · Cited by 10
SARSteer: Safeguarding Large Audio-Language Models via Safe-Ablated Refusal SteeringPreprint
Lin, W., Li, J., Xiong, H., & Liu, L. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2510.17633 · Cited by 10
Mechanistic Interpretability of Code Correctness in LLMs via Sparse AutoencodersPreprint
Tahimic, K., & Cheng, C. K. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2510.02917 · Cited by 10
Line of Sight: On Linear Representations in VLLMsPreprint
Rajaram, A., Schwettmann, S., Andreas, J., & Conmy, A. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2506.04706 · Cited by 10
GRP-Obliteration: Unaligning LLMs With a Single Unlabeled PromptPreprint
Russinovich, M., Cai, Y., Hines, K., Severi, G., Bullwinkel, B., & Salem, A. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2602.06258 · Cited by 11
Channel Location Constrains the Auditability of Subliminal LearningPreprint
Madl, T. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2606.22019 · Cited by 10
Improving Preference Extraction In LLMs By Identifying Latent Knowledge Through Classifying Probes
Maiya, S., Liu, Y., Debnath, R., & Korhonen, A. (2025) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2025.acl-long.444 · Cited by 9
In-Distribution Steering: Balancing Control and Coherence in Language Model Generation
Vogels, A., Wong, B., Choho, Y., Blangero, A., & Bhan, M. (2026) · Language Resources and Evaluation Conference (LREC) · doi:10.63317/4629fxavjicu · Cited by 9
Synthetic Eggs in Many Baskets: The Impact of Synthetic Data Diversity on LLM Fine-Tuning
Schaffelder, M., & Gatt, A. (2026) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2026.findings-acl.360 · Cited by 10
Representation Interventions Enable Lifelong Knowledge Memory Control in LLMs
Liu, X., Chen, S., Dong, X., Liu, Y., Zhao, X., Wang, H., Yan, Y., Chen, H., & Chen, Z. (2026) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2026.acl-long.246 · Cited by 9
Beyond I’m Sorry, I Can’t: Dissecting Large-Language-Model Refusal
Prakash, N., Jie, Y. W., Abdullah, A., Satapathy, R., Cambria, E., & Lee, R. K.-W. (2026) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v40i44.41119 · Cited by 9
LieCraft: A Multi-Agent Framework for Evaluating Deceptive Capabilities in Language Models
Olson, M., Ratzlaff, N., Hinck, M., Nguyen, T., Lal, V., Campbell, J., Stepputtis, S., & Tseng, S.-Y. (2026) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v40i44.41116 · Cited by 9
Burdisso, S., Baroudi, S., Labrak, Y., Grünert, D., Cyrta, P., Chen, Y., Madikeri, S. R., Villatoro-Tello, E., Marxer, R., & Motlíček, P. (2026) · Conference of the European Chapter of the Association for Computational Linguistics · doi:10.18653/v1/2026.eacl-demo.23 · Cited by 9
Haq, I., & Saldías-Fuentes, B. (2026) · ACM Conference on Fairness, Accountability, and Transparency (FAccT) · doi:10.1145/3805689.3812419 · Cited by 9
Understanding Emergent Misalignment via Feature Superposition Geometry
Minegishi, G., Furuta, H., Kojima, T., Iwasawa, Y., & Matsuo, Y. (2026) · Proceedings of the conference - Association for Computational Linguistics. Meeting · doi:10.18653/v1/2026.acl-long.1402 · Cited by 9
Safety Subspaces are Not Linearly Distinct: A Fine-Tuning Case Study
Ponkshe, K., Shah, S., Singhal, R., & Vepakomma, P. (2025) · International Conference on Learning Representations · doi:10.48550/arxiv.2505.14185 · Cited by 9
Language Models Are Implicitly Continuous
Marro, S., Evangelista, D., Huang, X. A., Malfa, E. L., Lombardi, M., & Wooldridge, M. (2025) · International Conference on Learning Representations · doi:10.48550/arxiv.2504.03933 · Cited by 10
Ham, S., Choi, Y., Yang, Y., Cho, S., Kim, Y., & Kim, C. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2506.07356 · Cited by 9
Steering LLMs for Formal Theorem ProvingPreprint
Kirtania, S., & Iyer, A. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2502.15507 · Cited by 9
Silenced Biases: The Dark Side LLMs Learned to RefusePreprint
Himelstein, R., LeVi, A., Youngmann, B., Nemcovsky, Y., & Mendelson, A. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2511.03369 · Cited by 9
Tracing Moral Foundations in Large Language ModelsPreprint
Yu, C., Yi, B., Karimi-Malekabadi, F., Abdurahman, S., Ye, J., Narayanan, S., Zhao, Y., & Dehghani, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2601.05437 · Cited by 10
Surgery: Mitigating Harmful Fine-Tuning for Large Language Models via Attention SinkPreprint
Liu, G., Lin, W., Huang, T., Mo, R., Mu, Q., Wang, X., & Shen, L. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2602.05228 · Cited by 9
Sparse Models, Sparse Safety: Unsafe Routes in Mixture-of-Experts LLMsPreprint
Jiang, Y., Huang, H., Li, M., Zhang, Y., Backes, M., & Zhang, Y. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2602.08621 · Cited by 10
Activation Steering via Generative Causal MediationPreprint
Sankaranarayanan, A., Zur, A., Geiger, A., & Hadfield-Menell, D. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2602.16080 · Cited by 10
Token-level Data Selection for Safe LLM Fine-tuning
Li, Y., Liu, Z., Li, Z., Lin, Z., & Zhang, J. (2026) · International Conference on Learning Representations · doi:10.48550/arxiv.2603.01185 · Cited by 9
Concept Heterogeneity-aware Representation SteeringPreprint
Abdullaev, L. U., Wong, N. Y. L., Lee, R. T. Z., Jiang, S., Nguyen, K., & Nguyen, T. M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2603.02237 · Cited by 9
LLMs Know They're Wrong and Agree Anyway: The Shared Sycophancy-Lying CircuitPreprint
Pandey, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2604.19117 · Cited by 9
Prompt-Activation Duality: Improving Activation Steering via Attention-Level InterventionsPreprint
Kang, D., Liu, Z., Ma, N., Huang, Y., Tan, Z., & Jiang, M. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.10664 · Cited by 9
FishBack: Pullback Fisher Geometry for Optimal Activation Steering in TransformersPreprint
Wang, S., Zhao, J., Cao, Q., Yao, H., & Shu, L. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2605.17231 · Cited by 9
Steering Robustness into World Action Models via Mechanistic Interpretability and Optimal ControlPreprint
Hong, J., Skifstad, J., Dai, Q., Chan, A., & Chou, G. (2026) · arXiv (Cornell University) · doi:10.48550/arxiv.2607.14943 · Cited by 9
Li, Y., Yi, X., Shi, D., de Melo, G., Wang, X., & Wang, L. (2025) · Findings of the Association for Computational Linguistics: ACL (Findings of ACL) · doi:10.18653/v1/2025.findings-acl.394 · Cited by 8
Turning the Spell Around: Lightweight Alignment Amplification via Rank-One Safety Injection
Shairah, H. A., Hammoud, H. A. A. K., Turkiyyah, G. M., & Ghanem, B. (2026) · Conference on Empirical Methods in Natural Language Processing (EMNLP) · doi:10.18653/2026.emnlp-main.75 · Cited by 8
Evading LLMs’ Safety Boundary with Adaptive Role-Play Jailbreaking
Wang, Z., Xie, W., Ma, S., Wang, E., & Wang, B. (2025) · Electronics · doi:10.3390/electronics14244808 · Cited by 8
Preserving Task-Relevant Information Under Linear Concept Removal
Holstege, F., Ravfogel, S., & Wouters, B. (2025) · neural information processing systems · doi:10.52202/085713-0907 · Cited by 8
ILRR: Inference-Time Steering Method for Masked Diffusion Language Models
Avrahami, E., & Nachmani, E. (2026) · International Joint Conference on Natural Language Processing and Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (IJCNLP-AACL) · doi:10.48550/arxiv.2601.21647 · Cited by 8
Silenced Biases: The Dark Side LLMs Learned to Refuse
Himelstein, R., LeVi, A., Youngmann, B., Nemcovsky, Y., & Mendelson, A. (2026) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v40i44.41078 · Cited by 8
Differentiated Directional Intervention: A Framework for Evading LLM Safety Alignment
Zhang, P., & Sun, P. (2026) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v40i44.41148 · Cited by 8
DAVSP: Safety Alignment for Large Vision-Language Models via Deep Aligned Visual Safety Prompt
Zhang, Y., Li, J., Cai, L., & Li, G. (2026) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v40i44.41149 · Cited by 8
Beyond Multiple Choice: Evaluating Steering Vectors for Summarization
Braun, J., Eickhoff, C., & Bahrainian, S. A. (2026) · Findings of the Association for Computational Linguistics: EACL · doi:10.18653/v1/2026.findings-eacl.200 · Cited by 8
The Unintended Trade-off of AI Alignment: Balancing Hallucination Mitigation and Safety in LLMs
Mahmoud, O., Khalil, A., Karimpanal, T. G., Semage, B. L., & Rana, S. (2026) · Findings of the Association for Computational Linguistics: EACL · doi:10.18653/v1/2026.findings-eacl.53 · Cited by 8
Analyzing Fine-Tuning Representation Shift for Multimodal LLMs Steering
Khayatan, P., Shukor, M., Parekh, J., Dapogny, A., & Cord, M. (2025) · IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv51701.2025.00213 · Cited by 8
Model Directions, Not Words: Mechanistic Topic Models Using Sparse Autoencoders
Zheng, C., Beltran-Velez, N., Karlekar, S., Shi, C., Nazaret, A., Mallik, A., Feder, A., & Blei, D. M. (2026) · Transactions of the Association for Computational Linguistics · doi:10.1162/tacl.a.797 · Cited by 8
Deep Model Merging: The Sister of Neural Network Interpretability -- A SurveyPreprint
Khan, A., Nief, T., Hudson, N., Sakarvadia, M., Grzenda, D., Ajith, A., Pettyjohn, J., Chard, K., & Foster, I. T. (2024) · arXiv (Cornell University) · doi:10.48550/arxiv.2410.12927 · Cited by 8
From Emergence to Control: Probing and Modulating Self-Reflection in Language Models
Zhu, X., Jiang, J., Khalili, M. M., & Zhu, Z. (2026) · Transactions on machine learning research · doi:10.48550/arxiv.2506.12217 · Cited by 8
Bridging the Knowledge-Prediction Gap in LLMs on Multiple-Choice QuestionsPreprint
Park, Y., Pyun, H., & Jo, Y. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2509.23782 · Cited by 8
AdversariaLLM: A Unified and Modular Toolbox for LLM Robustness ResearchPreprint
Beyer, T., Dornbusch, J., Steimle, J., Ladenburger, M., Schwinn, L., & Günnemann, S. (2025) · arXiv (Cornell University) · doi:10.48550/arxiv.2511.04316 · Cited by 8
Capability-Based Scaling Trends for LLM-Based Red-Teaming
Panfilov, A., Kassianik, P., Andriushchenko, M., & Geiping, J. (2026) · International Conference on Learning Representations · doi:10.48550/arxiv.2505.20162 · Cited by 8
Metadata
- Authors
- Arditi, A., Obeso, O., Syed, A., Paleka, D., Panickssery, N., Gurnee, W., & Nanda, N.
- ORCID
- Gurnee, Wes
- Venue
- Advances in Neural Information Processing Systems 37
- Published
- 2024
- Pages
- 136037-136083
- ISSN
- 1049-5258 (ISSN-L)
- OpenAlex
- W4415800641
- Cited by
- 70
- References
- 0
- Citations by year
- 2026: 61 · 2025: 9
- FWCI
- 386.89
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- neural information processing systems (publishedVersion)
- Topics
- Social Sciences › Arts and Humanities › Language and Linguistics › Language, Discourse, Communication Strategies
- Social Sciences › Arts and Humanities › Language and Linguistics › Linguistics and Language Analysis
- Physical Sciences › Computer Science › Artificial Intelligence › Speech and dialogue systems
- Keywords
- refusal, language models, large language models
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.