AI safety and alignment

Alignment, interpretability, red-teaming, bias and privacy: the research on what these systems do when they misbehave.

17,798 stories · RSS feed

arXiv AI

Sep 30

Trajectory Soup: Pushing the Compute-Scaling Frontier of LLM Mid-training via Diverse Trajectories

arXiv:2609.37169v1 Announce Type: cross Abstract: Mid-training equips pretrained large language models with specialized and reasoning capabilities, but the returns of this stage are bounded since add...

By Zhehao Huang, Changxin Tian, Qingyuan Yang, Kunlong Chen, Ziqi Liu, Zhiqiang Zhang, Xiaolin Huang, Jun Zhou

llms safety

arXiv Machine Learning

Sep 30

Byzantine-Robust Federated Representation Learning

arXiv:2609.36660v1 Announce Type: new Abstract: We study federated learning (FL) with adversarial clients, where the goal is to minimize the average loss of the honest (non-adversarial) clients witho...

By Leonardo F. Toso, James Anderson, Rafael Pinot, Nirupam Gupta

safety

arXiv Machine Learning

Sep 30

CHAIN: Calibrated LLM Forecasting via Causal-Temporal Hypergraph Inference

arXiv:2609.36689v1 Announce Type: new Abstract: Large language models have achieved significant progress in event forecasting, yet their probability outputs exhibit systematic calibration bias that v...

By Wenjin Liu, Chenxi Wang, Yue Lu, Zhe Cui, Haoran Luo

llms benchmarks safety

arXiv Machine Learning

Sep 30

Architecture Alignment With Sparse Priors in Tabular Foundation Models

arXiv:2609.36883v1 Announce Type: new Abstract: Tabular foundation models (TFMs) are increasingly popular because they deliver strong predictions on new datasets through in-context learning, without...

By Tianqi Zhao, Tianyi Zhuang, Shuo Duan, Guanyang Wang, Yan Shuo Tan, Qiong Zhang

llms safety

arXiv Machine Learning

Sep 30

A Model-Agnostic Physics-Guided Adapter for Few-Shot Transfer of Coastal Flood Prediction Models to Unseen Regions

arXiv:2609.37565v1 Announce Type: new Abstract: Deep learning surrogates can produce high-resolution coastal flood maps orders of magnitude faster than physics-based hydrodynamic simulators, yet tran...

By Bilal Hassan, Areg Karapetyan, Samer Madanat

fine-tuning safety

arXiv Machine Learning

Sep 30

Receptive-field-constrained stimulus optimization for human early and intermediate visual cortex

arXiv:2609.36391v1 Announce Type: cross Abstract: An ongoing challenge in sensory neuroscience is to characterize the feature dimensions encoded by cortical populations. Recent approaches probe featu...

By Junru Zhao, Hanfei Guo, Andrew Luo, Margaret M. Henderson

diffusion safety

arXiv Machine Learning

Sep 30

CAMEO: A Class-Activation-Mapped Equitable Overlay Framework for Fair and Robust Deep Learning-based Skin Condition Diagnosis

arXiv:2609.36400v1 Announce Type: cross Abstract: Deep learning classifiers for dermoscopic skin lesions often reach high in-distribution accuracy while quietly relying on spurious background cues su...

By Youssef Attia, Debasmita Mukherjee

safety

arXiv Machine Learning

Sep 30

Reactive Real-Time Flow Policies via Asynchronous Distribution Alignment

arXiv:2609.36540v1 Announce Type: cross Abstract: Generalist robot policies such as vision-language-action models (VLAs) have achieved remarkable generalization, but their inference delays can confli...

By Moritz Zoellner, Reece O'Mahoney, Ioannis Havoutis, Rohan Paleja

robotics efficiency multimodal safety

arXiv Machine Learning

Sep 30

Stochastic Decision Horizons for Survival-Based Constrained Reinforcement Learning

arXiv:2602.04599v3 Announce Type: replace Abstract: We propose stochastic decision horizons (SDH), a theoretically grounded framework for survival-based constrained reinforcement learning from per-st...

By Nikola Milosevic, Leonard Franz, Daniel Haeufle, Georg Martius, Nico Scherf, Pavel Kolev

reinforcement-learning safety