safety

General Frameworks for Conditional Two-Sample Testing

arXiv:2410.16636v2 Announce Type: replace-cross Abstract: We study the problem of conditional two-sample testing, which aims to determine whether two populations have the same distribution after accou

Generalized Distributional Alignment Games for Unbiased Answer-Level Fine-Tuning

DGX agent

arXiv:2605.02435v1 Announce Type: new Abstract: The Distributional Alignment Game framework provides a powerful variational perspective on Answer-Level Fine-Tuning (ALFT). However, standard algorithms

safetyarxiv-cs-lg

5 May 2026

Geometric and Spectral Alignment for Deep Neural Network I

arXiv:2605.02108v1 Announce Type: new Abstract: Deep residual architectures are modeled as products of near-identity Jacobians. This paper proves deterministic quotient-geometric estimates for singula

Geometric and Spectral Alignment for Deep Neural Network II

arXiv:2605.02111v1 Announce Type: new Abstract: This paper develops the angular and static-channel component of Geometric and Spectral Alignment for residual Jacobian chains. Starting from Cartan-coor

GETA-3DGS: Automatic Joint Structured Pruning and Quantization for 3D Gaussian Splatting

arXiv:2605.02086v1 Announce Type: new Abstract: 3D Gaussian splatting (3DGS) is a state-of-the-art representation for real-time photorealistic novel-view synthesis, yet a single high-fidelity scene ty

Good in Bad (GiB): Sifting Through End-user Demonstrations for Learning a Better Policy

arXiv:2605.01529v1 Announce Type: new Abstract: Imitation learning offers a promising framework for enabling robots to acquire diverse skills from human users. However, most imitation learning algorit

Gradient-Gated DPO: Stabilizing Preference Optimization in Language Models

arXiv:2605.02626v1 Announce Type: new Abstract: Preference optimization has become a central paradigm for aligning large language models with human feedback. Direct Preference Optimization (DPO) simpl

Green Energy Management for Sustainable Data Centers Using Deep Reinforcement Learning

arXiv:2507.21153v2 Announce Type: replace Abstract: The exponential growth of digital services has positioned data centers among the most energy-intensive infrastructures in the modern economy, raisin

HandelBot: Real-World Piano Playing via Fast Adaptation of Dexterous Robot Policies

arXiv:2603.12243v3 Announce Type: replace Abstract: Mastering dexterous manipulation with multi-fingered hands has been a grand challenge in robotics for decades. Despite its potential, the difficulty

Hazard-Aware Traffic Scene Graph Generation

arXiv:2603.03584v2 Announce Type: replace Abstract: Maintaining situational awareness in complex driving scenarios is challenging. It requires continuously prioritizing attention among extensive scene

HeteroRAG: A Heterogeneous Retrieval-Augmented Generation Framework for Medical Vision Language Tasks

arXiv:2508.12778v2 Announce Type: replace Abstract: Medical large vision-language Models (Med-LVLMs) have shown promise in clinical applications but suffer from factual inaccuracies and unreliable out

High entropy leads to symmetry equivariant policies in Dec-POMDPs

arXiv:2511.22581v3 Announce Type: replace Abstract: We prove that in any Dec-POMDP, sufficiently high entropy regularization ensures that the policy gradient flow with tabular softmax parametrization

How Can One Choose the Best CAM-Based Explainability Method for a CNN Model?

arXiv:2605.02007v1 Announce Type: cross Abstract: In recent years, several advances have been observed in Deep Learning with surprising results. Models in this area have been increasingly used in nume

HumanSplatHMR: Closing the Loop Between Human Mesh Recovery and Gaussian Splatting Avatar

arXiv:2605.02784v1 Announce Type: new Abstract: Accurately recovering human pose and appearance from video is an essential component of scene reconstruction, with applications to motion capture, motio

Hybrid Quantum Reinforcement Learning with QAOA for Improved Vehicle Routing Optimization

arXiv:2605.01574v1 Announce Type: new Abstract: Vehicle Routing Problem (VRP) is one of the most complex NP-hard combinatorial optimization problem in transportation and logistics that requires a dyna

Hydra-DP3: Frequency-Aware Right-Sizing of 3D Diffusion Policies for Visuomotor Control

arXiv:2605.01581v1 Announce Type: new Abstract: Diffusion-based visuomotor policies perform well in robotic manipulation, yet current methods still inherit image-generation-style decoders and multi-st

Hyp2Former: Hierarchy-Aware Hyperbolic Embeddings for Open-Set Panoptic Segmentation

arXiv:2605.02580v1 Announce Type: new Abstract: Recognizing unknown objects is crucial for safety-critical applications such as autonomous driving and robotics. Open-Set Panoptic Segmentation (OPS) ai

I am old enough to remember when people used to believe every word of Sam’s bullshit (and to call me a “hater” for doubting him).

Gary Marcus reflects on changing public perception of Sam Altman, noting that skepticism toward Altman's claims was once dismissed as hatred but has become more mainstream. The post suggests a shift i

I predict this will be the most embarrassing cover in the history of this magazine. This is like putting Enron executives on your cover, or …

I predict this will be the most embarrassing cover in the history of this magazine. This is like putting Enron executives on your cover, or doing a gauzy special report on Bernie Madoff. Sam Altman is

Implicature in Interaction: Understanding Implicature Improves Alignment in Human-LLM Interaction

arXiv:2510.25426v2 Announce Type: replace Abstract: The rapid advancement of Large Language Models (LLMs) is positioning language at the core of human-computer interaction (HCI). We argue that advanci

Important new development. AI company employees have an enormous amount of power — far more than they realize. Absent legislation, AI co wor…

Important new development. AI company employees have an enormous amount of power — far more than they realize. Absent legislation, AI co worker power is one of the key levers to shaping what the indus

Improving Model Safety by Targeted Error Correction

arXiv:2605.02544v1 Announce Type: cross Abstract: The widespread adoption of machine learning in critical applications demands techniques to mitigate high-consequence errors. Our method utilizes a dua

In vividly explaining how he deceived Musk about his commitment to the nonprofit, without a trace of remorse, OpenAI’s Brockman has done Mus…

In vividly explaining how he deceived Musk about his commitment to the nonprofit, without a trace of remorse, OpenAI’s Brockman has done Musk’s counsel quite a service. Brockman’s also been pretty gre

Injecting Distributional Awareness into MLLMs via Reinforcement Learning for Deep Imbalanced Regression

arXiv:2605.01402v1 Announce Type: new Abstract: Multimodal large language models (MLLMs) struggle with numerical regression under long-tailed target distributions. Token-level supervised fine-tuning (

Investigating Anthropometric Fidelity in SAM 3D Body

arXiv:2601.06035v2 Announce Type: replace-cross Abstract: The release of SAM 3D Body is a recent development in human mesh recovery, demonstrating improved performance in producing clean, topologicall

IPS: In-Prompt Process Supervision for Short Video Content Moderation

arXiv:2412.15251v3 Announce Type: replace Abstract: Multimodal large language models (MLLMs) are effective at capturing the semantics of short video content; however, they often fail to attend to the

🆕 @katiemiller has started following @GaryMarcus

Katie Miller began following Gary Marcus on X (formerly Twitter). Gary Marcus is a cognitive scientist and AI researcher known for his public commentary on artificial intelligence and technology polic

Knowledge-Based Design Requirements for Generative Social Robots in Higher Education

arXiv:2602.12873v4 Announce Type: replace-cross Abstract: Generative social robots (GSRs) powered by large language models enable adaptive, conversational tutoring but also introduce risks such as mis

safetyarxiv-cs-ai

Lateral String Stability for Vehicle Platoons: Formulation, Definition, and Analysis

arXiv:2605.01731v1 Announce Type: new Abstract: Platooning of connected and automated vehicles provides significant benefits in terms of energy efficiency, traffic throughput, and, most critically, sa

Learning to Act Through Contact: A Unified View of Multi-Task Robot Learning

arXiv:2510.03599v2 Announce Type: replace Abstract: We present a unified framework for multi-task locomotion and manipulation policy learning grounded in a contact-explicit representation. Instead of

Less is More: Geometric Unlearning for LLMs with Minimal Data Disclosure

arXiv:2605.01735v1 Announce Type: new Abstract: As large language models (LLMs) are increasingly deployed in real-world systems, they must support post-hoc removal of specific content to meet privacy

Less Precise Can Be More Reliable: A Systematic Evaluation of Quantization's Impact on VLMs Beyond Accuracy

arXiv:2509.21173v5 Announce Type: replace Abstract: Vision-Language Models (VLMs) such as CLIP have revolutionized zero-shot classification and safety-critical tasks, including Out-of-Distribution (OO

Linking spatial biology and clinical histology via Haiku

arXiv:2605.00925v1 Announce Type: cross Abstract: Integrating molecular, morphological, and clinical data is essential for basic and translational biomedical research, yet systematic frameworks for jo

LLM-Augmented Semantic Steering of Text Embedding Projection Spaces

arXiv:2605.01957v1 Announce Type: cross Abstract: Low-dimensional projections of text embeddings support visual analysis of document collections, but their spatial organization may not reflect the rel

LLM-Based Agentic Negotiation for 6G: Addressing Uncertainty Neglect and Tail-Event Risk

arXiv:2511.19175v2 Announce Type: replace-cross Abstract: A critical barrier to the trustworthiness of sixth-generation (6G) agentic autonomous networks is the uncertainty neglect bias; a cognitive te

safetyarxiv-cs-ai

LLM-VA: Resolving the Jailbreak-Overrefusal Trade-off via Vector Alignment

arXiv:2601.19487v2 Announce Type: replace Abstract: Safety-aligned LLMs suffer from two failure modes: jailbreak (answering harmful inputs) and over-refusal (declining benign queries). Existing vector

Logit-Gap Steering: A Forward-Pass Diagnostic for Alignment Robustness

arXiv:2506.24056v2 Announce Type: replace-cross Abstract: RLHF-style alignment trains language models to refuse unsafe requests, but how much operational margin does this refusal rest on? We introduce

Low-Latency Video Anonymization for Crowd Anomaly Detection: Privacy Versus Performance

arXiv:2410.18717v2 Announce Type: replace Abstract: Recent advancements in artificial intelligence hold ample potential for monitoring applications using surveillance cameras. However, concerns about

LVLM-Aided Alignment of Task-Specific Vision Models

arXiv:2512.21985v2 Announce Type: replace Abstract: In high-stakes domains, small task-specific vision models are crucial due to their low computational requirements and the availability of numerous m

Machine Learning Enhanced Laser Spectroscopy for Multi-Species Gas Detection in Complex and Harsh Environments

arXiv:2605.01306v1 Announce Type: cross Abstract: Laser absorption spectroscopy (LAS) is a well-established technique for non-intrusive measurement of gas species in combustion and atmospheric environ

MAD-OPD: Breaking the Ceiling in On-Policy Distillation via Multi-Agent Debate

arXiv:2605.01347v1 Announce Type: new Abstract: On-policy distillation (OPD) trains a student on its own trajectories under token-level teacher supervision, but existing methods are capped by a single

Major new class action lawsuit accuses Meta of copyright infringement around AI training. It says they trained on pirated books, and that AI…

Major new class action lawsuit accuses Meta of copyright infringement around AI training. It says they trained on pirated books, and that AI books flooding the market demonstrates market harm. These l

Manifold-Constrained Adversarial Training for Long-Tailed Robustness via Geometric Alignment

arXiv:2605.02183v1 Announce Type: new Abstract: Adversarial training is effective on balanced datasets, but its robustness degrades under longtailed class distributions, where tail classes suffer high

Mean Testing under Truncation beyond Gaussian

arXiv:2605.01335v1 Announce Type: cross Abstract: We characterize the fundamental limits of high-dimensional mean testing under arbitrary truncation, where samples are drawn from the conditional distr

Meta says it will expand Instagram teen account safeguards to 27 EU countries, and plans to roll them out on Facebook in the US, ahead of the UK and EU in June (Foo Yun Chee/Reuters)

Foo Yun Chee / Reuters: Meta says it will expand Instagram teen account safeguards to 27 EU countries, and plans to roll them out on Facebook in the US, ahead of the UK and EU in June — Meta Platforms

safetytechmeme

Minimizing Collateral Damage in Activation Steering

arXiv:2605.01167v1 Announce Type: new Abstract: Activation steering is a method for controlling Large Language Model (LLM) behavior by intervening in its internal representations to increase the align

MIRA: A Score for Conditional Distribution Accuracy and Model Comparison

arXiv:2605.02014v1 Announce Type: cross Abstract: We introduce Mira, a sample-based score for assessing the accuracy of a candidate conditional distribution using only joint samples from the true data

Mitigating Misalignment Contagion by Steering with Implicit Traits

arXiv:2605.02751v1 Announce Type: cross Abstract: Language models (LMs) are increasingly used in high-stakes, multi-agent settings, where following instructions and maintaining value alignment are cri