AI safety and alignment

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

17,788 stories · RSS feed

arXiv Computer Vision

Oct 27

When Pointing Becomes Profiling: Social Bias and Refusal in Text-Steerable Visual Encoders

arXiv:2610.35017v1 Announce Type: new Abstract: Text-steerable visual encoders condition a frozen Vision Transformer on a natural-language prompt and point at the matching image region, producing pro...

By Seyedarmin Azizi, Parsa Razmara, Sajjad Ghiasvand, Ardalan Aryashad, Tina Khezresmaeilzadeh

llms computer-vision benchmarks safety

arXiv Computer Vision

Oct 27

Reading the Registers of Honorius III without Manual Line Transcription: Forced-Aligned HTR, End to End, Modular or Hybrid

arXiv:2610.35085v1 Announce Type: new Abstract: Unlocking medieval archives at scale is bottlenecked by the cost of line-level diplomatic transcriptions and manual abbreviation expansion. Focusing on...

By Tamas Kovacs, Daniel Luger

computer-vision safety

arXiv Computer Vision

Oct 27

Uncertainty Evaluation to Support Safety of the Intended Functionality Analysis for Determining the Applicability of Performance Targets Across Triggering Conditions in ML-Based LiDAR Object Detection

arXiv:2610.35092v1 Announce Type: new Abstract: For safety of the intended functionality (SOTIF) analysis according to ISO 21448, a machine learning object detector for light detection and ranging (L...

By Milin Patel, Rolf Jung

computer-vision safety

arXiv Computer Vision

Oct 27

Low-rank MERA Decomposition of 3D OCT Image Volume for Compression, Denoising, Feature Extraction and Diagnosis of Retinal Diseases

arXiv:2610.35271v1 Announce Type: new Abstract: Optical coherence tomography (OCT) is a non-invasive imaging modality used for visualization of retinal tissue. However, OCT volumes are affected by sp...

By Ivica Kopriva, Fei Shi, Marija \v{S}tanfel, Qin Peng, Jianhao Bao, Gaowei Li, Xinjian Chen

computer-vision safety

arXiv Computer Vision

Oct 27

S$^2$-RL: Sample-Set Dual Reinforcement Learning for Generative Semantic Segmentation Dataset Distillation

arXiv:2610.35491v1 Announce Type: new Abstract: While dataset distillation has witnessed significant advances in image classification, its extension to dense prediction tasks such as semantic segment...

By Haoyu Wang, Fei Zhou, Qingqing Qiu, Lei Zhang, Wei Wei, Chen Ding

diffusion computer-vision reinforcement-learning fine-tuning efficiency benchmarks safety

arXiv Computer Vision

Oct 27

Beyond Distribution Matching: Self-Supervised Representation Forcing for Few-Step Video Generation

arXiv:2610.35594v1 Announce Type: new Abstract: Few-step distillation is essential for deploying modern video diffusion models. Among existing approaches, Distribution Matching Distillation (DMD) has...

By Junyi Chen, Haiyu Zhang, Zhoujie Fu, Tengfei Wang, Chunchao Guo

diffusion efficiency safety

arXiv Computer Vision

Oct 27

Spectrally optimized implicit neural representations mitigate acquisition-related image quality trade-offs in intracranial 4D Flow MRI

arXiv:2610.33124v1 Announce Type: cross Abstract: 4D Flow MRI promises opportunities for non-invasive mapping of neurovascular hemodynamics, but its clinical utility faces trade-offs between resoluti...

By Oliver Welin Odeback, Nerea Gonz\'alez-Aranceta, Sebasti\'an Jofre, Javier Bisbal, Edward Ferdian, Alistair A. Young, Pia Callmer, Sameer A. Ansari, Mia Bonini, David Nordsletten, Patrick Winter, Susanne Schnell, Tobias Granberg, Alexander Fyrdahl, David Marlevi

safety

arXiv Computer Vision

Oct 27

Stable Dimensionality Reduction via Isometric Centroid-Encoder

arXiv:2610.33872v1 Announce Type: cross Abstract: Many supervised dimensionality reduction methods promote intra-class compactness and inter-class separation but do not explicitly constrain class geo...

By Tomojit Ghosh, Siva Rama Krishna Prasad Changala, Prathyusha Kanakamalla, Harigovind Harikumar

rag diffusion safety

arXiv Computer Vision

Oct 27

MMGNet: Multi-Modal 6-DoF Grasp Synthesis for Cluttered Scenes

arXiv:2610.34893v1 Announce Type: cross Abstract: Grasping is a foundational capability in robotic manipulation. Scene-level 6-DoF grasp synthesis relies on point clouds for metric geometry, yet real...

By Rui Cao, Jia-Hui Pan, Yichuan Li, Yunfan Yu, Chi-Wing Fu, Jiangliu Wang, Yun-Hui Liu

robotics multimodal benchmarks safety

arXiv Computer Vision

Oct 27

MedVTok: A General-Purpose Medical Visual Tokenizer

arXiv:2610.35842v1 Announce Type: cross Abstract: Multimodal medical AI requires a common visual interface for encoding heterogeneous imaging evidence for patient condition modeling. However, existin...

By Chenglong Ma, Yuanfeng Ji, Junzhi Ning, Jiyao Liu, Yujie Wei, Shaohao Rui, Chaoyang Zhang, Wenjie Li, Wenhao Tang, Cheng Tang, Wei Li, Shujian Gao, Yujie Zhang, Ying Chen, Zilong Li, Zongyuan Ge, Diping Song, Jin Ye, Ming Hu, Junjun He, Hongming Shan

computer-vision nlp multimodal benchmarks safety

arXiv Computer Vision

Oct 27

On the Theory of Bias Tuning in Event Cameras

arXiv:2501.18788v4 Announce Type: replace Abstract: This paper lays the foundation of a theory for bias tuning in neuromorphic cameras, a novel sensing technology also known as "event cameras". We be...

By David El-Chai Ben-Ezra, Daniel Brisk, Adar Tal

safety

arXiv Computer Vision

Oct 27

DAOS: A Multimodal In-cabin Behavior Monitoring with Driver Action-Object Synergy Dataset

arXiv:2601.11990v2 Announce Type: replace Abstract: Multimodal action understanding requires complementary sensory observations to be connected with structured semantic knowledge. This is particularl...

By Yiming Li, Dan Lin, Wenfei Liang, Tianyi Liu, Chen Cai, Wenqian Wang, Kim-Hui Yap, Liang Zhao, Bingbing Li

llms multimodal safety

arXiv Computer Vision

Oct 27

Resolving Primitive-Sharing Ambiguity in Long-Tailed TLS-Based Industrial MEP Point Cloud Segmentation via Spatial Context Constraints

arXiv:2601.19128v3 Announce Type: replace Abstract: In terrestrial laser scanning (TLS)-based mechanical, electrical, and plumbing (MEP) point cloud segmentation, safety-critical components such as r...

By Chao Yin, Qing Han, Zhiwei Hou, Yue Liu, Anjin Dai, Hongda Hu, Ji Yang, Wei Yao

computer-vision safety

arXiv Computer Vision

Oct 27

Can a Teenager Fool an AI? Evaluating Low-Cost Cosmetic Attacks on Age Estimation Systems

arXiv:2602.19539v2 Announce Type: replace Abstract: Age estimation systems are increasingly deployed as gatekeepers for age-restricted online content, yet their robustness to cosmetic modifications h...

By Xingyu Shen, Tommy Duong, Xiaodong An, Zengqi Zhao, Zebang Hu, Haoyu Hu, Ziyou Wang, Finn Guo, Simiao Ren

llms multimodal benchmarks safety

arXiv Computer Vision

Oct 27

Know3D: Prompting 3D Generation with Knowledge from Vision-Language Models

arXiv:2603.22782v3 Announce Type: replace Abstract: Recent advances in 3D generation have improved the fidelity and geometric details of synthesized 3D assets. However, due to the inherent ambiguity...

By Wenyue Chen, Wenjue Chen, Peng Li, Qinghe Wang, Xu Jia, Heliang Zheng, Rongfei Jia, Yuan Liu, Ronggang Wang

llms diffusion multimodal safety

arXiv Computer Vision

Oct 27

A Synthetic Eye Movement Dataset for Script Reading Detection: Real Trajectory Replay on a 3D Simulator

arXiv:2604.05475v2 Announce Type: replace Abstract: Large vision-language models have achieved remarkable capabilities by training on massive internet-scale data, yet a fundamental asymmetry persists...

By Kidus Zewde, Yuchen Zhou, Dennis Ng, Neo Tiangratanakul, Tommy Duong, Ankit Raj, Yuxin Zhang, Xingyu Shen, Simiao Ren

llms robotics multimodal safety

arXiv Computer Vision

Oct 27

Cranio-Diff: Diffusion-based Cross-domain Craniofacial Reconstruction with 2D X-ray Skull Guidance and Structural Identity Constraints

arXiv:2606.09699v2 Announce Type: replace Abstract: The identification of unknown human remains is a persistent humanitarian and forensic challenge. In forensic investigations, it becomes even more c...

By Ravi Shankar Prasad, Naresh Gurjar, Shashank Baghel, Chirag, Dinesh Singh

diffusion safety

arXiv Computer Vision

Oct 27

ScenePercept: Scenario-Conditioned Perceptual Assessment of Real-World User-Generated Content Images

arXiv:2610.17661v2 Announce Type: replace Abstract: The rapid development of mobile imaging and social platforms has led to explosive growth in user-generated content images, making their systematic...

By Shuhan Xia, Yu Chen, Xiang Wang, Shuai Li, Peipie Li, Zekun Li

benchmarks safety

arXiv Computer Vision

Oct 27

RoboEvolve: Co-Evolving Planner-Simulator for Robotic Manipulation with Limited Data

arXiv:2605.13775v2 Announce Type: replace-cross Abstract: The scalability of robotic manipulation is fundamentally bottlenecked by the scarcity of task-aligned physical interaction data. While vision...

By Harold Haodong Chen, Sirui Chen, Yingjie Xu, Wenhang Ge, Ying-Cong Chen

llms agents diffusion reinforcement-learning robotics multimodal safety

arXiv Statistics ML

Oct 27

Projection-Based Federated Learning on Compact Submanifolds with Geometry-Independent Differential Privacy

arXiv:2610.33712v1 Announce Type: new Abstract: Federated learning trains a shared model across clients without exchanging their data, yet the updates they transmit still leak information about indiv...

By Thibault Pautrel, Florent Bouchard, Guillaume Ginolhac, Ammar Mian, Aur\'elien Bellet

safety

Get the daily briefing