← Back to all news

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 Computer Vision
Sep 22

G6D: Geometric Learning-Free RGB-D 6D Pose Solver for Robotic Manipulation

G6D is a learning‑free, geometry‑driven RGB‑D 6D pose solver designed for robotic manipulation. It generates pose hypotheses via template‑based geometric matching and refines them using silhouette and depth consistency, requiring only an RGB‑D observation, an object mask, camera intrinsics, and a CAD model. The method offers adjustable accuracy‑computation trade‑offs, can run on CPU without GPUs, and has shown strong performance on LineMOD and BOP19 datasets, as well as in real‑world pick‑and‑place experiments.

By Yixuan Liang (Tsinghua University), William Chen (Sapient Intelligence), Yunan Wang (Tsinghua University), Jizhou Yan (Tsinghua University), Zhao Jin (Tsinghua University), Changling Liu (Sapient Intelligence), Chuxiong Hu (Tsinghua University)
computer-visionroboticssafety
More like this →
arXiv Computer Vision
Sep 22

A$^2$Safe: Counterfactual Evidence-Aligned Adaptive Agent Collaboration for Safe and Effective Visual Question Answering

A$^2$Safe is a framework for safe and effective Visual Question Answering that aligns counterfactual evidence with adaptive agent collaboration. It uses a Grounded Safety Evidence Board to make safety decisions explicit, enforcing invariance to safety‑irrelevant changes while allowing appropriate transitions when risk‑critical evidence changes. The system achieves a 95.72 SIUO safety score, reduces benign refusals on MOSSBench to 14.67%, and maintains a 78.34 average VQA score with 27.8% token overhead.

By Quanxing Xu, Ling Zhou, Xian Zhong, Jinyu Tian, Xiaohua Huang, Rubing Huang, Chia-Wen Lin
llmsagentsnlpmultimodalsafety
More like this →
arXiv Computer Vision
Sep 22

Classifier-Free Guidance in Flow Matching: Non-Autonomous Potentials, Overshoot, and Posterior-Mean Control

The paper studies classifier‑free guidance (CFG) in Flow Matching, showing that strong guidance can distort the generated distribution by shifting the mean and concentrating trajectories. By interpreting Flow Matching as a time‑varying gradient flow, the authors explain how CFG reshapes the underlying potential and propose a training‑free method, Posterior‑Mean‑Capped CFG (PMC‑CFG), that adaptively limits guidance to the strongest feasible level. Experiments on synthetic and large‑scale image‑generation tasks demonstrate that PMC‑CFG reduces distortion and concentration while improving the alignment–diversity trade‑off, especially when nominal guidance is large.

By Jishen Peng, Zheng Ma
agentsdiffusionbenchmarkssafety
More like this →
arXiv Computer Vision
Sep 22

0.5\%>100\%: Bidirectional Reciprocal Learning for Referring Image Segmentation

The paper introduces Bidirectional Reciprocal Learning (BRL), a parameter‑efficient fine‑tuning framework for referring image segmentation that operates on frozen vision foundation models. BRL employs two lightweight adapters—Reciprocal Attention Adapter (RAA) for token‑level cross‑modal attention and Reciprocal Gate Adapter (RGA) for channel‑level gating—to enable hierarchical, bidirectional information flow between vision and language. Experiments on RefCOCO, RefCOCO+, and RefCOCOg show that BRL outperforms existing methods while updating fewer than 0.5% of backbone parameters.

By Xiaoqiang Lu, Licheng Jiao, Lingling Li, Yuting Yang, Long Sun, Wenping Ma, Xu Liu, Fang Liu
computer-visionfine-tuningmultimodalbenchmarkssafety
More like this →
arXiv Machine Learning
Sep 22

SafeTune: A Unified Faithful Library for Auditing and Repairing Safety Drift in Fine-Tuned LLMs

SafeTune is a source‑available library that consolidates four safety‑intervention paradigms—post‑hoc weight recovery, safety‑constrained fine‑tuning, gradient‑based unlearning, and inference‑time steering—into a single, configuration‑driven workflow. It offers shared interpretability, evaluation, and deployment tools, and its modular registry allows easy addition of new methods, benchmarks, judges, models, and fine‑tuning domains. The authors demonstrate SafeTune with controlled comparisons and case studies in finance and medical deployments, showing how it characterizes safety drift, evaluates interventions on refusal‑behavior and capability metrics, and supports calibrated or layered mitigation.

By Pratinav Seth, Saisab Sadhu, Anshul Kaushal, Vinay Kumar Sankarapu
llmsfine-tuningbenchmarkssafety
More like this →
arXiv Machine Learning
Sep 22

Autonomous Model Lifecycle Management for Digital Twin-Based Manufacturing Control

The paper introduces a closed‑loop cyber‑physical system for autonomous model lifecycle management in automotive manufacturing, deployed since 2023. It manages paired physics and reinforcement‑learning models, selecting the best candidate through competitive retraining cycles and a Conductor orchestrator that handles plant‑wide inventories and fallback controls. The system incorporates an operator‑trust gate that rejects 23% of policies that deviate from established practice, achieving 28‑45% process stability improvements with no safety incidents.

By Zhengyang (Cissy), Gu, Thomas Cook, Fredaljohn Rohrbaugh, Joseph E. Hernandez, Chris Couch
agentsreinforcement-learningsafety
More like this →
arXiv Machine Learning
Sep 22

Counterfactual Tool Ranking under Utility, Cost, and Privilege Constraints

The paper introduces a counterfactual tool ranking framework that accounts for authority, historical support, and estimation nuances. Using eleven enterprise-inspired tools, synthetic and real-world experiments on the Berkeley Function Calling Leaderboard, the study compares direct regression and doubly robust (DR) methods, finding that DR performs better in shifted environments while direct regression excels in linear settings. The authors also evaluate Qwen2.5 models on held-out tasks, analyze policy differences under missing support, and present a falsifiable evaluation method with publicly available evidence.

By Jiapeng Li
agentsbenchmarkssafety
More like this →
arXiv Machine Learning
Sep 22

EAVer: Long-Form Factuality Verification as an End-to-End Agentic Policy

arXiv:2609.22223v1 Announce Type: cross Abstract: Long-form factuality verification is commonly implemented as a static decompose-search-verify pipeline, with separately prompted modules processing c...

By Kening Zheng, Aoying Zheng, Zhigang Chang, Yazhi Guo, Miaotian Guo, Qingwei Zong, Xianhai Xie, Weiqiang Jin, Chengze Li, Hanrong Zhang, Jie Yang, Wei-Chieh Huang, Lingzhe Zhang, Liancheng Fang, Xin Zou, Hanqian Li, Jiahao Huo, Yibo Yan, Zizhuang Deng, Lei Miao, Wei Guo, Haihong Tang, Bo Zheng, Philip S. Yu
llmsagentsreinforcement-learningbenchmarkssafety
More like this →