English
Related papers

Related papers: Persistent Pre-Training Poisoning of LLMs

200 papers

Deep neural networks are vulnerable to backdoor attacks, a type of adversarial attack that poisons the training data to manipulate the behavior of models trained on such data. Clean-label attacks are a more stealthy form of backdoor attacks…

Machine Learning · Computer Science 2024-07-17 Quang H. Nguyen , Nguyen Ngoc-Hieu , The-Anh Ta , Thanh Nguyen-Tang , Kok-Seng Wong , Hoang Thanh-Tung , Khoa D. Doan

Pretraining NLP models with variants of Masked Language Model (MLM) objectives has recently led to a significant improvements on many tasks. This paper examines the benefits of pretrained models as a function of the number of training…

Computation and Language · Computer Science 2020-06-17 Sinong Wang , Madian Khabsa , Hao Ma

Recent generative large language models (LLMs) show remarkable performance in non-English languages, but when prompted in those languages they tend to express higher harmful social biases and toxicity levels. Prior work has shown that…

Computation and Language · Computer Science 2025-06-03 Vera Neplenbroek , Arianna Bisazza , Raquel Fernández

While large language models have achieved remarkable performance on various code generation benchmarks, there have been growing concerns regarding potential contamination of these benchmarks as they may be leaked into pretraining and…

Software Engineering · Computer Science 2024-03-11 Martin Riddell , Ansong Ni , Arman Cohan

Adversaries can embed backdoors in deep learning models by introducing backdoor poison samples into training datasets. In this work, we investigate how to detect such poison samples to mitigate the threat of backdoor attacks. First, we…

Machine Learning · Computer Science 2023-06-21 Xiangyu Qi , Tinghao Xie , Jiachen T. Wang , Tong Wu , Saeed Mahloujifar , Prateek Mittal

We investigate fingerprints in pretraining datasets for large language models (LLMs) through dataset classification experiments. Building on prior work demonstrating the existence of fingerprints or biases in popular computer vision…

Machine Learning · Computer Science 2025-12-02 Youssef Mansour , Reinhard Heckel

Training AI models in cybersecurity with help of vast datasets offers significant opportunities to mimic real-world behaviors effectively. However, challenges like data drift and scarcity of labelled data lead to frequent updates of models…

Machine Learning · Computer Science 2026-02-04 Saurabh Anand , Shubham Malaviya , Manish Shukla , Sachin Lodha

We present a controlled study of how dataset contamination interacts with the post-training stages now standard in large language model training pipelines. Starting from clean checkpoints of Qwen2.5 (0.5B/1.5B) and Gemma3 (1B/4B), we inject…

Machine Learning · Computer Science 2026-01-13 Muhammed Yusuf Kocyigit , Caglar Yildirim

Large Language Models (LLMs) are often trained with safety guards intended to prevent harmful text generation. However, such safety training can be removed by fine-tuning the LLM on harmful datasets. While this emerging threat (harmful…

Computation and Language · Computer Science 2024-10-04 Domenic Rosati , Jan Wehner , Kai Williams , Łukasz Bartoszcze , Jan Batzner , Hassan Sajjad , Frank Rudzicz

Chain-of-Thought (CoT) reasoning has emerged as a powerful technique for enhancing large language models' capabilities by generating intermediate reasoning steps for complex tasks. A common practice for equipping LLMs with reasoning is to…

Cryptography and Security · Computer Science 2026-01-29 Harsh Chaudhari , Ethan Rathbun , Hanna Foerster , Jamie Hayes , Matthew Jagielski , Milad Nasr , Ilia Shumailov , Alina Oprea

The advent of Large Language Models (LLMs) has revolutionized various applications by providing advanced natural language processing capabilities. However, this innovation introduces new cybersecurity challenges. This paper explores the…

Cryptography and Security · Computer Science 2024-06-18 Stephen Burabari Tete

Large language model (LLM) systems increasingly power everyday AI applications such as chatbots, computer-use assistants, and autonomous robots, where performance often depends on manually well-crafted prompts. LLM-based prompt optimizers…

Machine Learning · Computer Science 2026-01-14 Andrew Zhao , Reshmi Ghosh , Vitor Carvalho , Emily Lawton , Keegan Hines , Gao Huang , Jack W. Stokes

Large language models (LLMs) learn undesirable properties during pretraining, including dangerous knowledge and toxic text generation. Just as post-training uses different objectives to shape different behaviors, we argue that unlearning…

Computation and Language · Computer Science 2026-05-27 Berk Atil , Vipul Gupta , Rebecca J. Passonneau

It is commonplace to produce application-specific models by fine-tuning large pre-trained models using a small bespoke dataset. The widespread availability of foundation model checkpoints on the web poses considerable risks, including the…

Cryptography and Security · Computer Science 2024-04-02 Yuxin Wen , Leo Marchyok , Sanghyun Hong , Jonas Geiping , Tom Goldstein , Nicholas Carlini

We introduce camouflaged data poisoning attacks, a new attack vector that arises in the context of machine unlearning and other settings when model retraining may be induced. An adversary first adds a few carefully crafted points to the…

Machine Learning · Computer Science 2024-08-02 Jimmy Z. Di , Jack Douglas , Jayadev Acharya , Gautam Kamath , Ayush Sekhari

AI-based code generators have gained a fundamental role in assisting developers in writing software starting from natural language (NL). However, since these large language models are trained on massive volumes of data collected from…

Cryptography and Security · Computer Science 2024-03-12 Cristina Improta

As modern neural machine translation (NMT) systems have been widely deployed, their security vulnerabilities require close scrutiny. Most recently, NMT systems have been found vulnerable to targeted attacks which cause them to produce…

Computation and Language · Computer Science 2021-02-16 Chang Xu , Jun Wang , Yuqing Tang , Francisco Guzman , Benjamin I. P. Rubinstein , Trevor Cohn

Large Language Models (LLMs) have advanced Graph Neural Networks (GNNs) by enriching node representations with semantic features, giving rise to LLM-enhanced GNNs that achieve notable performance gains. However, the robustness of these…

Machine Learning · Computer Science 2026-03-30 Yuhang Ma , Jie Wang , Zheng Yan

Large Language Models (LLMs) can generate content that is as persuasive as human-written text and appear capable of selectively producing deceptive outputs. These capabilities raise concerns about potential misuse and unintended…

Computation and Language · Computer Science 2024-12-24 Cameron R. Jones , Benjamin K. Bergen

Large Language Models (LLMs) have increasingly become pivotal in content generation with notable societal impact. These models hold the potential to generate content that could be deemed harmful.Efforts to mitigate this risk include…

Computation and Language · Computer Science 2024-08-20 Kexin Chen , Yi Liu , Dongxia Wang , Jiaying Chen , Wenhai Wang