CORTEXA
← Browse
arxivcs.AIcs.CL2026-07-16

Pretraining Data Can Be Poisoned through Computational Propaganda

Victoria Graf, Hannaneh Hajishirzi, Noah A. Smith, David Kohlbrenner, Kyle Lo

Poisoning pretraining data can introduce harmful behaviors to LMs that are difficult to detect and mitigate. Prior work on poisoning pretraining data has largely exploited established data sources such as Wikipedia, which do not represent the large scale and heterogeneity typical of pretraining corpora, and has ignored the interaction between poisoned data and data curation pipelines. We demonstrate that poisoning attacks on pretraining data are feasible beyond this limited setting through an existing web-scale content injection mechanism: public discussion interfaces. Additionally, to measure whether malicious content is included after web crawling and data curation, we introduce HalfLife, a novel analysis for estimating adversarial content inclusion in web-crawl based LM training data. We use HalfLife to explore the feasibility of poisoning pretraining corpora at web scale through open discussion interfaces. Our analysis demonstrates the importance of estimating whether poison injections are included in pretraining data, and establishes third-party webpage content as a possible vector for attacking language model pretraining.

View free PDFSource page

Related papers

arxivcs.ROcs.AIcs.CLcs.CRcs.LG2026-07-05

!Imperio, smolVLA: The Implications of Data Poisoning on Open Source Robotics

Stefan Bühler, Mark Schutera

This work establishes that trigger-word data poisoning of vision language action models is practical, while at the same time the open-source robotics ecosystem holds trust assumptions about community contributions. A few poisoned samples can silently embed a backdoor that disable…

View free PDFSource page
arxivcs.LGcs.AIcs.CL2026-07-17

Understanding Reasoning from Pretraining to Post-Training

Jingyan Shen, Ang Li, Salman Rahman, Yifan Sun, Micah Goldblum, Matus Telgarsky, et al.

Reinforcement learning (RL) has become central to improving large language models (LLMs) on complex reasoning tasks, yet RL post-training is largely studied in isolation from the pretraining that precedes it. As a result, two basic questions remain open: (1) how do pretraining ch…

View free PDFSource page
arxivcs.CLcs.AI2026-07-11

Polarization Detection: A Hybrid Approach with AfroXLMR-Social and DeBERTa for Low- and High-Resource Settings

Muhammad Abdullahi Said

The rapid proliferation of online polarization threatens social cohesion, necessitating robust automated detection systems that operate effectively across diverse linguistic contexts. This paper presents our system description for the POLAR Shared Task 2026, focusing on the detec…

View free PDFSource page
arxivcs.SDcs.AIcs.CL2026-06-27

Clustering Unsupervised Representations as Defense against Poisoning Attacks on Speech Commands Classification System

Thomas Thebaud, Sonal Joshi, Henry Li, Martin Sustek, Jesus Villalba, Sanjeev Khudanpur, et al.

Poisoning attacks entail attackers intentionally tampering with training data. In this paper, we consider a dirty-label poisoning attack scenario on a speech commands classification system. The threat model assumes that certain utterances from one of the classes (source class) ar…

View free PDFSource page
arxivcs.CVcs.AIcs.CL2026-07-01

MindEdit-Bench: Benchmarking Object-Level Counterfactual Spatial Reasoning in VLMs from In-the-Wild Photos

Leyuan Yu, Xiao Tang, Minghao Liu, Xinyuan Li, Xiaokai Bai, Sheng Zhou, et al.

Benchmarks for vision-language models (VLMs) mostly test observational spatial reasoning: models describe relations already visible in the input. Existing what-if tasks typically vary the observer while keeping the scene fixed. Can VLMs instead predict the consequences of hypothe…

View free PDFSource page