Cover Image

Winter Soldier: Backdooring Language Models at Pre-Training with Indirect Data Poisoning

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Bouaziz, Wassim, Videau, Mathurin, Usunier, Nicolas, El-Mhamdi, El-Mahdi
Format:	Preprint
Published:	2025
Subjects:	Cryptography and Security Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Inverting Gradient Attacks Makes Powerful Data Poisoning
by: Bouaziz, Wassim, et al.
Published: (2024)

Data Taggants: Dataset Ownership Verification via Harmless Targeted Data Poisoning
by: Bouaziz, Wassim, et al.
Published: (2024)

Targeted Data Poisoning for Black-Box Audio Datasets Ownership Verification
by: Bouaziz, Wassim, et al.
Published: (2025)

Approaching the Harm of Gradient Attacks While Only Flipping Labels
by: El-Kabid, Abdessamad, et al.
Published: (2025)

Privacy Backdoors: Enhancing Membership Inference through Poisoning Pre-trained Models
by: Wen, Yuxin, et al.
Published: (2024)

Hiding Backdoors within Event Sequence Data via Poisoning Attacks
by: Ermilova, Alina, et al.
Published: (2023)

Backdoor Learning Curves: Explaining Backdoor Poisoning Beyond Influence Functions
by: Cinà, Antonio Emanuele, et al.
Published: (2021)

Provable Robustness of (Graph) Neural Networks Against Data Poisoning and Backdoor Attacks
by: Gosch, Lukas, et al.
Published: (2024)

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

State Backdoor: Towards Stealthy Real-world Poisoning Attack on Vision-Language-Action Model in State Space
by: Guo, Ji, et al.
Published: (2026)

Indiscriminate Data Poisoning Attacks on Pre-trained Feature Extractors
by: Lu, Yiwei, et al.
Published: (2024)

Layer of Truth: Probing Belief Shifts under Continual Pre-Training Poisoning
by: Churina, Svetlana, et al.
Published: (2025)

The 'Sure' Trap: Multi-Scale Poisoning Analysis of Stealthy Compliance-Only Backdoors in Fine-Tuned Large Language Models
by: Tan, Yuting, et al.
Published: (2025)

Concealing Backdoor Model Updates in Federated Learning by Trigger-Optimized Data Poisoning
by: Zhang, Yujie, et al.
Published: (2024)

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

Protecting Copyright of Medical Pre-trained Language Models: Training-Free Backdoor Model Watermarking
by: Kong, Cong, et al.
Published: (2024)

Graph-Aware Stealthy Poison-Text Backdoors for Text-Attributed Graphs
by: Luo, Qi, et al.
Published: (2026)

PoisonedParrot: Subtle Data Poisoning Attacks to Elicit Copyright-Infringing Content from Large Language Models
by: Panaitescu-Liess, Michael-Andrei, et al.
Published: (2025)

Model Supply Chain Poisoning: Backdooring Pre-trained Models via Embedding Indistinguishability
by: Wang, Hao, et al.
Published: (2024)

SleeperNets: Universal Backdoor Poisoning Attacks Against Reinforcement Learning Agents
by: Rathbun, Ethan, et al.
Published: (2024)

Secure Transfer Learning: Training Clean Models Against Backdoor in (Both) Pre-trained Encoders and Downstream Datasets
by: Zhang, Yechao, et al.
Published: (2025)

Poison with Style: A Practical Poisoning Attack on Code Large Language Models
by: Tran, Khang, et al.
Published: (2026)

Beyond Training-time Poisoning: Component-level and Post-training Backdoors in Deep Reinforcement Learning
by: Vyas, Sanyam, et al.
Published: (2025)

Backdooring Masked Diffusion Language Models
by: Cao, Daniel Yiming, et al.
Published: (2026)

Multi-Trigger Poisoning Amplifies Backdoor Vulnerabilities in LLMs
by: Sivapiromrat, Sanhanat, et al.
Published: (2025)

SteganoBackdoor: Stealthy and Data-Efficient Backdoor Attacks on Language Models
by: Xue, Eric, et al.
Published: (2025)

Fusing Pruned and Backdoored Models: Optimal Transport-based Data-free Backdoor Mitigation
by: Lin, Weilin, et al.
Published: (2024)

Poisoning Web-Scale Training Datasets is Practical
by: Carlini, Nicholas, et al.
Published: (2023)

Backdoor Federated Learning by Poisoning Backdoor-Critical Layers
by: Zhuang, Haomin, et al.
Published: (2023)

Backdoor Attacks on Decentralised Post-Training
by: Ersoy, Oğuzhan, et al.
Published: (2026)

ToxicTextCLIP: Text-Based Poisoning and Backdoor Attacks on CLIP Pre-training
by: Yao, Xin, et al.
Published: (2025)

PCAP-Backdoor: Backdoor Poisoning Generator for Network Traffic in CPS/IoT Environments
by: Chathoth, Ajesh Koyatan, et al.
Published: (2025)

Adaptive and Robust Data Poisoning Detection and Sanitization in Wearable IoT Systems using Large Language Models
by: Mithsara, W. K. M, et al.
Published: (2025)

Privacy Backdoors: Stealing Data with Corrupted Pretrained Models
by: Feng, Shanglun, et al.
Published: (2024)

Concept-ROT: Poisoning Concepts in Large Language Models with Model Editing
by: Grimes, Keltin, et al.
Published: (2024)

Provable Watermarking for Data Poisoning Attacks
by: Zhu, Yifan, et al.
Published: (2025)

CorruptEncoder: Data Poisoning based Backdoor Attacks to Contrastive Learning
by: Zhang, Jinghuai, et al.
Published: (2022)

A Systematic Review of Poisoning Attacks Against Large Language Models
by: Fendley, Neil, et al.
Published: (2025)

Self-Purification Mitigates Backdoors in Multimodal Diffusion Language Models
by: Wan, Guangnian, et al.
Published: (2026)

Unelicitable Backdoors in Language Models via Cryptographic Transformer Circuits
by: Draguns, Andis, et al.
Published: (2024)