English
Related papers

Related papers: Subliminal Corruption: Mechanisms, Thresholds, and…

200 papers

We study subliminal learning, a surprising phenomenon where language models transmit behavioral traits via semantically unrelated data. In our main experiments, a "teacher" model with some trait T (such as liking owls or being misaligned)…

Machine Learning · Computer Science 2025-07-22 Alex Cloud , Minh Le , James Chua , Jan Betley , Anna Sztyber-Betley , Jacob Hilton , Samuel Marks , Owain Evans

Recent work on subliminal learning demonstrates that language models can transmit semantic traits through data that is semantically unrelated to those traits. However, it remains unclear whether behavioral traits can transfer in agentic…

Artificial Intelligence · Computer Science 2026-04-20 Jacob Dang , Brian Y. Xie , Omar G. Younis

Previous research has shown that fine-tuning language models on general tasks enhance their underlying mechanisms. However, the impact of fine-tuning on poisoned data and the resulting changes in these mechanisms are poorly understood. This…

Machine Learning · Computer Science 2025-03-05 Vishnu Kabir Chhabra , Ding Zhu , Mohammad Mahdi Khalili

Subliminal learning describes a student language model inheriting a behavioral bias by fine-tuning on seemingly innocuous data generated by a biased teacher model. Prior work has begun to characterize this phenomenon but leaves open…

Computation and Language · Computer Science 2026-04-29 George Morgulis , John Hewitt

As machine learning models are increasingly deployed in safety-critical domains, visual explanation techniques have become essential tools for supporting transparency. In this work, we reveal a new class of attacks that compromise model…

Computer Vision and Pattern Recognition · Computer Science 2025-11-19 Farhin Farhad Riya , Shahinul Hoque , Jinyuan Stella Sun , Olivera Kotevska

Corruption is notoriously widespread in data collection. Despite extensive research, the existing literature predominantly focuses on specific settings and learning scenarios, lacking a unified view of corruption modelization and…

Machine Learning · Computer Science 2026-05-19 Laura Iacovissi , Nan Lu , Robert C. Williamson

This paper proposes a training data augmentation pipeline that combines synthetic image data with neural style transfer in order to address the vulnerability of deep vision models to common corruptions. We show that although applying style…

Computer Vision and Pattern Recognition · Computer Science 2025-12-22 Georg Siedel , Rojan Regmi , Abhirami Anand , Weijia Shao , Silvia Vock , Andrey Morozov

We develop and study new adversarial perturbations that enable an attacker to gain control over decisions in generic Artificial Intelligence (AI) systems including deep learning neural networks. In contrast to adversarial data modification,…

Cryptography and Security · Computer Science 2023-12-07 Ivan Y. Tyukin , Desmond J. Higham , Alexander Bastounis , Eliyas Woldegeorgis , Alexander N. Gorban

Understanding how language models generalize behaviors from their training to a broader deployment distribution is an important problem in AI safety. Betley et al. discovered that fine-tuning GPT-4o on intentionally insecure code causes…

Indiscriminate data poisoning attacks aim to decrease a model's test accuracy by injecting a small amount of corrupted training data. Despite significant interest, existing attacks remain relatively ineffective against modern machine…

Machine Learning · Computer Science 2023-06-07 Yiwei Lu , Gautam Kamath , Yaoliang Yu

Fine-tuning large language models on narrow data with harmful content produces broadly misaligned behavior on unrelated prompts, a phenomenon known as emergent misalignment. We propose that emergent misalignment involves persona-model…

Computation and Language · Computer Science 2026-05-26 Davi Bastos Costa , Renato Vicente

Data corruption, including missing and noisy data, poses significant challenges in real-world machine learning. This study investigates the effects of data corruption on model performance and explores strategies to mitigate these effects…

Machine Learning · Computer Science 2025-05-22 Qi Liu , Wanjing Ma

As large language models (LLMs) become more powerful and are deployed more autonomously, it will be increasingly important to prevent them from causing harmful outcomes. Researchers have investigated a variety of safety techniques for this…

Machine Learning · Computer Science 2024-07-24 Ryan Greenblatt , Buck Shlegeris , Kshitij Sachan , Fabien Roger

Understanding AI systems' inner workings is critical for ensuring value alignment and safety. This review explores mechanistic interpretability: reverse engineering the computational mechanisms and representations learned by neural networks…

Artificial Intelligence · Computer Science 2024-08-27 Leonard Bereska , Efstratios Gavves

Previous research has shown that LLMs finetuned on malicious or incorrect completions within narrow domains (e.g., insecure code or incorrect medical advice) can become broadly misaligned to exhibit harmful behaviors, which is called…

Computation and Language · Computer Science 2026-01-21 Xuhao Hu , Peng Wang , Xiaoya Lu , Dongrui Liu , Xuanjing Huang , Jing Shao

We study the extent to which standard machine learning algorithms rely on exchangeability and independence of data by introducing a monotone adversarial corruption model. In this model, an adversary, upon looking at a "clean" i.i.d.…

Machine Learning · Computer Science 2026-01-06 Kasper Green Larsen , Chirag Pabbaraju , Abhishek Shetty

Synthetic corruptions gathered into a benchmark are frequently used to measure neural network robustness to distribution shifts. However, robustness to synthetic corruption benchmarks is not always predictive of robustness to distribution…

Computer Vision and Pattern Recognition · Computer Science 2021-11-19 Alfred Laugros , Alice Caplier , Matthieu Ospici

Reward hacking--where agents exploit flaws in imperfect reward functions rather than performing tasks as intended--poses risks for AI alignment. Reward hacking has been observed in real training runs, with coding agents learning to…

Artificial Intelligence · Computer Science 2025-08-26 Mia Taylor , James Chua , Jan Betley , Johannes Treutlein , Owain Evans

Even when prompts and parameters are secured, transformer language models remain vulnerable because their key-value (KV) cache during inference constitutes an overlooked attack surface. This paper introduces Malicious Token Injection (MTI),…

Cryptography and Security · Computer Science 2026-02-03 Elias Hossain , Swayamjit Saha , Somshubhra Roy , Ravi Prasad

In supervised learning one wishes to identify a pattern present in a joint distribution $P$, of instances, label pairs, by providing a function $f$ from instances to labels that has low risk $\mathbb{E}_{P}\ell(y,f(x))$. To do so, the…

Machine Learning · Statistics 2015-07-07 Brendan van Rooyen , Robert C. Williamson
‹ Prev 1 2 3 10 Next ›