Cover Image

Towards Understanding the Robustness of Sparse Autoencoders

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Saiyed, Ahson, Sadiekh, Sabrina, Agarwal, Chirag
Format:	Preprint
Published:	2026
Subjects:	Machine Learning Artificial Intelligence Computation and Language Cryptography and Security
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Certifying LLM Safety against Adversarial Prompting
by: Kumar, Aounon, et al.
Published: (2023)

SAEs $\textit{Can}$ Improve Unlearning: Dynamic Sparse Autoencoder Guardrails for Precision Unlearning in LLMs
by: Muhamed, Aashiq, et al.
Published: (2025)

Towards Building a Robust Toxicity Predictor
by: Bespalov, Dmitriy, et al.
Published: (2024)

Towards Understanding the Fragility of Multilingual LLMs against Fine-Tuning Attacks
by: Poppi, Samuele, et al.
Published: (2024)

Towards Robust Knowledge Unlearning: An Adversarial Framework for Assessing and Improving Unlearning Robustness in Large Language Models
by: Yuan, Hongbang, et al.
Published: (2024)

Logicbreaks: A Framework for Understanding Subversion of Rule-based Inference
by: Xue, Anton, et al.
Published: (2024)

On Adversarial Robustness of Language Models in Transfer Learning
by: Turbal, Bohdan, et al.
Published: (2024)

Directional Embedding Smoothing for Robust Vision Language Models
by: Wang, Ye, et al.
Published: (2026)

FIT to Forget: Robust Continual Unlearning for Large Language Models
by: Xu, Xiaoyu, et al.
Published: (2026)

OBLIVIATE: Robust and Practical Machine Unlearning for Large Language Models
by: Xu, Xiaoyu, et al.
Published: (2025)

Probing the Robustness of Large Language Models Safety to Latent Perturbations
by: Gu, Tianle, et al.
Published: (2025)

Was it Slander? Towards Exact Inversion of Generative Language Models
by: Skapars, Adrians, et al.
Published: (2024)

PostMark: A Robust Blackbox Watermark for Large Language Models
by: Chang, Yapei, et al.
Published: (2024)

SELF: A Robust Singular Value and Eigenvalue Approach for LLM Fingerprinting
by: Zhang, Hanxiu, et al.
Published: (2025)

Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM
by: Cao, Bochuan, et al.
Published: (2023)

Less is More: Understanding Word-level Textual Adversarial Attack via n-gram Frequency Descend
by: Lu, Ning, et al.
Published: (2023)

Tracing the Dynamics of Refusal: Exploiting Latent Refusal Trajectories for Robust Jailbreak Detection
by: Hu, Xulin, et al.
Published: (2026)

AliMark: Enhancing Robustness of Sentence-Level Watermarking Against Text Paraphrasing
by: Li, Yuexin, et al.
Published: (2026)

Tuning without Peeking: Provable Generalization Bounds and Robust LLM Post-Training
by: Labiad, Ismail, et al.
Published: (2025)

Does Low Rank Adaptation Lead to Lower Robustness against Training-Time Attacks?
by: Liang, Zi, et al.
Published: (2025)

MOCHA: Are Code Language Models Robust Against Multi-Turn Malicious Coding Prompts?
by: Wahed, Muntasir, et al.
Published: (2025)

SVIP: Towards Verifiable Inference of Open-source Large Language Models
by: Sun, Yifan, et al.
Published: (2024)

Sparse Autoencoders are Capable LLM Jailbreak Mitigators
by: Assogba, Yannick, et al.
Published: (2026)

Polarity-Aware Probing for Quantifying Latent Alignment in Language Models
by: Sadiekh, Sabrina, et al.
Published: (2025)

Toward a Safer Web: Multilingual Multi-Agent LLMs for Mitigating Adversarial Misinformation Attacks
by: Aldahoul, Nouar, et al.
Published: (2025)

Special Characters Attack: Toward Scalable Training Data Extraction From Large Language Models
by: Bai, Yang, et al.
Published: (2024)

Memories Retrieved from Many Paths: A Multi-Prefix Framework for Robust Detection of Training Data Leakage in Large Language Models
by: Dang, Trung Cuong, et al.
Published: (2025)

Sparse Tokens Suffice: Jailbreaking Audio Language Models via Token-Aware Gradient Optimization
by: Fang, Zheng, et al.
Published: (2026)

Enhancing Variational Autoencoders with Smooth Robust Latent Encoding
by: Lee, Hyomin, et al.
Published: (2025)

Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red Teaming
by: Sharma, Mrinank, et al.
Published: (2025)

Exploring the Robustness of In-Context Learning with Noisy Labels
by: Cheng, Chen, et al.
Published: (2024)

Adversarial Intent is a Latent Variable: Stateful Trust Inference for Securing Multimodal Agentic RAG
by: Singh, Inderjeet, et al.
Published: (2026)

Jailbreak Foundry: From Papers to Runnable Attacks for Reproducible Benchmarking
by: Fang, Zhicheng, et al.
Published: (2026)

Exposing the Systematic Vulnerability of Open-Weight Models to Prefill Attacks
by: Struppek, Lukas, et al.
Published: (2026)

Exposing LLM Safety Gaps Through Mathematical Encoding:New Attacks and Systematic Analysis
by: Zhang, Haoyu, et al.
Published: (2026)

LLM Ghostbusters: Surgical Hallucination Suppression via Adaptive Unlearning
by: Spracklen, Joseph, et al.
Published: (2026)

PIArena: A Platform for Prompt Injection Evaluation
by: Geng, Runpeng, et al.
Published: (2026)

Cross-Session Threats in AI Agents: Benchmark, Evaluation, and Algorithms
by: Azarafrooz, Ari
Published: (2026)

Adaptive Instruction Composition for Automated LLM Red-Teaming
by: Zymet, Jesse, et al.
Published: (2026)

MANATEE: Inference-Time Lightweight Diffusion Based Safety Defense for LLMs
by: Kan, Chun Yan Ryan, et al.
Published: (2026)