Imagen de Portada

Stateless Yet Not Forgetful: Implicit Memory as a Hidden Channel in LLMs

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Salem, Ahmed, Paverd, Andrew, Abdelnabi, Sahar
Formato:	Preprint
Publicado:	2026
Materias:	Machine Learning Artificial Intelligence Cryptography and Security
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

No More, No Less: Task Alignment in Terminal Agents
por: Mavali, Sina, et al.
Publicado: (2026)

Counter-Samples: A Stateless Strategy to Neutralize Black Box Adversarial Attacks
por: Bokobza, Roey, et al.
Publicado: (2024)

A Cognac Shot To Forget Bad Memories: Corrective Unlearning for Graph Neural Networks
por: Kolipaka, Varshita, et al.
Publicado: (2024)

Get my drift? Catching LLM Task Drift with Activation Deltas
por: Abdelnabi, Sahar, et al.
Publicado: (2024)

Learning to Forget using Hypernetworks
por: Rangel, Jose Miguel Lara, et al.
Publicado: (2024)

Obliviate: Efficient Unmemorization for Protecting Intellectual Property in Large Language Models
por: Russinovich, Mark, et al.
Publicado: (2025)

Reasoning Introduces New Poisoning Attacks Yet Makes Them More Complicated
por: Foerster, Hanna, et al.
Publicado: (2025)

Continual Learning with Strategic Selection and Forgetting for Network Intrusion Detection
por: Zhang, Xinchen, et al.
Publicado: (2024)

The Erasure Illusion: Stress-Testing the Generalization of LLM Forgetting Evaluation
por: Jia, Hengrui, et al.
Publicado: (2025)

Measuring Security Without Fooling Ourselves: Why Benchmarking Agents Is Hard
por: Abdelnabi, Sahar, et al.
Publicado: (2026)

ACU: Analytic Continual Unlearning for Efficient and Exact Forgetting with Privacy Preservation
por: Tang, Jianheng, et al.
Publicado: (2025)

Data Unlearning Beyond Uniform Forgetting via Diffusion Time and Frequency Selection
por: Park, Jinseong, et al.
Publicado: (2025)

Forget to Flourish: Leveraging Machine-Unlearning on Pretrained Language Models for Privacy Leakage
por: Rashid, Md Rafi Ur, et al.
Publicado: (2024)

BadSampler: Harnessing the Power of Catastrophic Forgetting to Poison Byzantine-robust Federated Learning
por: Liu, Yi, et al.
Publicado: (2024)

Securing AI Agents with Information-Flow Control
por: Costa, Manuel, et al.
Publicado: (2025)

Skill-Inject: Measuring Agent Vulnerability to Skill File Attacks
por: Schmotz, David, et al.
Publicado: (2026)

Embedding Hidden Adversarial Capabilities in Pre-Trained Diffusion Models
por: Beerens, Lucas, et al.
Publicado: (2025)

Frequency maps reveal the correlation between Adversarial Attacks and Implicit Bias
por: Basile, Lorenzo, et al.
Publicado: (2023)

Unveiling the Backdoor Mechanism Hidden Behind Catastrophic Overfitting in Fast Adversarial Training
por: Zhao, Mengnan, et al.
Publicado: (2026)

Predictive Auditing of Hidden Tokens in LLM APIs via Reasoning Length Estimation
por: Wang, Ziyao, et al.
Publicado: (2025)

Hidden in Memory: Sleeper Memory Poisoning in LLM Agents
por: Pulipaka, Sidharth, et al.
Publicado: (2026)

Enhancing Continual Learning for Software Vulnerability Prediction: Addressing Catastrophic Forgetting via Hybrid-Confidence-Aware Selective Replay for Temporal LLM Fine-Tuning
por: Dou, Xuhui, et al.
Publicado: (2026)

Enhancing Vulnerability Reports with Automated and Augmented Description Summarization
por: Althebeiti, Hattan, et al.
Publicado: (2025)

ConVerse: Benchmarking Contextual Safety in Agent-to-Agent Conversations
por: Gomaa, Amr, et al.
Publicado: (2025)

When Evaluation Becomes a Side Channel: Regime Leakage and Structural Mitigations for Alignment Assessment
por: Santos-Grueiro, Igor
Publicado: (2026)

Automated Consistency Analysis of LLMs
por: Patwardhan, Aditya, et al.
Publicado: (2025)

Channel-Level Semantic Perturbations: Unlearnable Examples for Diverse Training Paradigms
por: Wang, Bo, et al.
Publicado: (2026)

FLARE: A Wireless Side-Channel Fingerprinting Attack on Federated Learning
por: Shuvo, Md Nahid Hasan, et al.
Publicado: (2025)

Large-scale online deanonymization with LLMs
por: Lermen, Simon, et al.
Publicado: (2026)

Scaling Trends for Data Poisoning in LLMs
por: Bowen, Dillon, et al.
Publicado: (2024)

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

Leveraging RAG for Training-Free Alignment of LLMs
por: Halloran, John T.
Publicado: (2026)

Injecting Universal Jailbreak Backdoors into LLMs in Minutes
por: Chen, Zhuowei, et al.
Publicado: (2025)

Shape and Substance: Dual-Layer Side-Channel Attacks on Local Vision-Language Models
por: Hadad, Eyal, et al.
Publicado: (2026)

TimeGuard: Channel-wise Pool Training for Backdoor Defense in Time Series Forecasting
por: Nguyen, Quang Duc, et al.
Publicado: (2026)

DPrivBench: Benchmarking LLMs' Reasoning for Differential Privacy
por: Wang, Erchi, et al.
Publicado: (2026)

RL-Finetuned LLMs for Privacy-Preserving Synthetic Rewriting
por: Shi, Zhan, et al.
Publicado: (2025)

Fast Exact Unlearning for In-Context Learning Data for LLMs
por: Muresanu, Andrei I., et al.
Publicado: (2024)

Agentic Misalignment: How LLMs Could Be Insider Threats
por: Lynch, Aengus, et al.
Publicado: (2025)

Attention Tracker: Detecting Prompt Injection Attacks in LLMs
por: Hung, Kuo-Han, et al.
Publicado: (2024)