Imagen de Portada

Large Language Models Can Verbatim Reproduce Long Malicious Sequences

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Lin, Sharon, Krishnamurthy, Dvijotham, Hayes, Jamie, Shi, Chongyang, Shumailov, Ilia, Song, Shuang
Formato:	Preprint
Publicado:	2025
Materias:	Machine Learning
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Buffer Overflow in Mixture of Experts
por: Hayes, Jamie, et al.
Publicado: (2024)

Interpreting the Repeated Token Phenomenon in Large Language Models
por: Yona, Itay, et al.
Publicado: (2025)

Beyond Slow Signs in High-fidelity Model Extraction
por: Foerster, Hanna, et al.
Publicado: (2024)

Measuring memorization in RLHF for code completion
por: Pappu, Aneesh, et al.
Publicado: (2024)

Cascading Adversarial Bias from Injection to Distillation in Language Models
por: Chaudhari, Harsh, et al.
Publicado: (2025)

Stealing User Prompts from Mixture of Experts
por: Yona, Itay, et al.
Publicado: (2024)

Inexact Unlearning Needs More Careful Evaluations to Avoid a False Sense of Privacy
por: Hayes, Jamie, et al.
Publicado: (2024)

Soft Instruction De-escalation Defense
por: Walter, Nils Philipp, et al.
Publicado: (2025)

Lessons from Defending Gemini Against Indirect Prompt Injections
por: Shi, Chongyang, et al.
Publicado: (2025)

Quantamination: Dynamic Quantization Leaks Your Data Across the Batch
por: Foerster, Hanna, et al.
Publicado: (2026)

Demystifying Verbatim Memorization in Large Language Models
por: Huang, Jing, et al.
Publicado: (2024)

Thought-Transfer: Indirect Targeted Poisoning Attacks on Chain-of-Thought Reasoning Models
por: Chaudhari, Harsh, et al.
Publicado: (2026)

Locking Machine Learning Models into Hardware
por: Clifford, Eleanor, et al.
Publicado: (2024)

Fairness Feedback Loops: Training on Synthetic Data Amplifies Bias
por: Wyllie, Sierra, et al.
Publicado: (2024)

Achieving the Tightest Relaxation of Sigmoids for Formal Verification
por: Chevalier, Samuel, et al.
Publicado: (2024)

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

Norm-Bounded Low-Rank Adaptation
por: Wang, Ruigang, et al.
Publicado: (2025)

Monotone, Bi-Lipschitz, and Polyak-Lojasiewicz Networks
por: Wang, Ruigang, et al.
Publicado: (2024)

SEA: Shareable and Explainable Attribution for Query-based Black-box Attacks
por: Gao, Yue, et al.
Publicado: (2023)

Machine Learning needs Better Randomness Standards: Randomised Smoothing and PRNG-based attacks
por: Dahiya, Pranav, et al.
Publicado: (2023)

Keeping up with dynamic attackers: Certifying robustness to adaptive online data poisoning
por: Bose, Avinandan, et al.
Publicado: (2025)

Architectural Backdoors for Within-Batch Data Stealing and Model Inference Manipulation
por: Küchler, Nicolas, et al.
Publicado: (2025)

Measuring memorization in language models via probabilistic extraction
por: Hayes, Jamie, et al.
Publicado: (2024)

Honeyval: A Comprehensive Evaluation Framework for LLM-powered HTTP Honeypots
por: Vero, Mark, et al.
Publicado: (2026)

ceLLMate: Sandboxing Browser AI Agents
por: Meng, Luoxi, et al.
Publicado: (2025)

ParaPO: Aligning Language Models to Reduce Verbatim Reproduction of Pre-training Data
por: Chen, Tong, et al.
Publicado: (2025)

Beyond Labeling Oracles: What does it mean to steal ML models?
por: Shafran, Avital, et al.
Publicado: (2023)

Watermarking Needs Input Repetition Masking
por: Khachaturov, David, et al.
Publicado: (2025)

CrisperWhisper: Accurate Timestamps on Verbatim Speech Transcriptions
por: Wagner, Laurin, et al.
Publicado: (2024)

Language Models May Verbatim Complete Text They Were Not Explicitly Trained On
por: Liu, Ken Ziyu, et al.
Publicado: (2025)

The Attacker Moves Second: Stronger Adaptive Attacks Bypass Defenses Against Llm Jailbreaks and Prompt Injections
por: Nasr, Milad, et al.
Publicado: (2025)

Hardware and Software Platform Inference
por: Zhang, Cheng, et al.
Publicado: (2024)

Revisiting Block-based Quantisation: What is Important for Sub-8-bit LLM Inference?
por: Zhang, Cheng, et al.
Publicado: (2023)

UnUnlearning: Unlearning is not sufficient for content regulation in advanced generative AI
por: Shumailov, Ilia, et al.
Publicado: (2024)

Machine Learning Models Have a Supply Chain Problem
por: Meiklejohn, Sarah, et al.
Publicado: (2025)

Confidence-aware Reward Optimization for Fine-tuning Text-to-Image Models
por: Kim, Kyuyoung, et al.
Publicado: (2024)

Beyond Laplace and Gaussian: Exploring the Generalized Gaussian Mechanism for Private Machine Learning
por: Rinberg, Roy, et al.
Publicado: (2025)

ImpNet: Imperceptible and blackbox-undetectable backdoors in compiled neural networks
por: Clifford, Eleanor, et al.
Publicado: (2022)

When Vision Fails: Text Attacks Against ViT and OCR
por: Boucher, Nicholas, et al.
Publicado: (2023)

No, of Course I Can! Deeper Fine-Tuning Attacks That Bypass Token-Level Safety Mechanisms
por: Kazdan, Joshua, et al.
Publicado: (2025)