Safety & alignment papers
What AI researchers are reading: the papers the Hugging Face community picks each day, and new arXiv submissions in the directions we follow. Titles and abstracts are arXiv’s; every paper links to arXiv.
Mechanistic Interpretability Needs a Logic of Inference
Zhimin Hu, Chen Jiang, Hanbo Xie et al.
Neural networks are increasingly studied not merely as predictive systems, but as complex systems whose internal computations are themselves objects of scientific inquiry. Recent progress in mechanistic interpretability (MI) has made it possible to identify neurons, features, circuits, population structures, and…
Causal Concept Explanations for Deep Neural Models
Josh Rountree, Pulkit Verma, Oswin So et al.
Concept-based explanations in neural models are important for the development, certification, and adoption of autonomous systems. Yet, interpretability remains correlational. Without methods to guide and quantify causal relationships between explanations, a concept explanation can be provided without being…
PSL-RGHA: Parallel Split Learning with Reference Gradient Half-Space Alignment for Mitigating Gradient Divergence
Han Li, Zheng Lin, Miao Yang et al.
The escalating complexity of deep neural networks hinders the deployment of federated learning (FL) on resource-constrained devices, especially for multimedia applications such as on-device image classification and speech recognition. Parallel split learning (PSL) has emerged as a promising solution by offloading a…
Are Causal Neuron Circuits Unique and Transferable?
Ömer Tekin Yavuz
Interpretability demonstrations often show that intervening on a group of neurons changes a model's answer. A recent example removes an 18-neuron group in Qwen3-32B and the model answers "Simba" instead of "Scar" to a two-hop Lion King riddle. Such a demonstration establishes that the group matters, but not how…
Where Does Alignment Emerge and What Does It Mean? A Closer Look at the Platonic Representation Hypothesis
Julia Herbst, Simon Roschmann, Quentin Bouniot et al.
The Platonic Representation Hypothesis (PRH) suggests that neural networks are converging toward similar representation spaces. As language models become more capable, their representations are aligning with those of vision models. However, recent work has questioned whether this trend continues to hold for newer…
SWAM: Scaling World Action Models with Cross-Modal Alignment
Yu Zhou, Xinyuan Xia, Xuankun Yang et al.
World action models jointly predict video and actions conditioned on language. Existing studies focus primarily on architectures and training mechanisms, typically treating language-video alignment acquired through video pretraining as an established foundation. However, introducing action learning may impair this…
Shortcuts and Boundaries: Verifiable RL Misalignment and Contextual Safety Learning
Xinjie Shen
Reinforcement learning (RL) can improve an agent's task capability while reinforcing behavior that violates contextual execution requirements. We study this reward-driven misalignment, including reward hacking, through verifiable reward--context conflicts, where compliant task completion remains feasible while some…