safety

When Choices Become Risks: Safety Failures of Large Language Models under Multiple-Choice Constraints

arXiv:2604.16916v1 Announce Type: new Abstract: Safety alignment in large language models (LLMs) is primarily evaluated under open-ended generation, where models can mitigate risk by refusing to respo

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Where Do Self-Supervised Speech Models Become Unfair?

DGX agent

arXiv:2604.18249v1 Announce Type: new Abstract: Speech encoder models are known to model members of some speaker groups (SGs) better than others. However, there has been little work in establishing wh

safetyarxiv-cs-cl

21 Apr 2026

Who Gets the Kidney? Human-AI Alignment, Indecision, and Moral Values

arXiv:2506.00079v2 Announce Type: replace-cross Abstract: The rapid integration of Large Language Models (LLMs) in high-stakes decision-making -- such as allocating scarce resources like donor organs

Why Agents Compromise Safety Under Pressure

arXiv:2603.14975v2 Announce Type: replace-cross Abstract: Large Language Model agents deployed in complex environments frequently encounter a conflict between maximizing goal achievement and adhering

Why AI Readiness Is an Organizational Learning Problem, Not a Technology Purchase

arXiv:2604.16369v1 Announce Type: cross Abstract: Global corporate AI investment reached $252.3 billion in 2024, yet only 6% of firms report significant earnings impact. This article argues that AI pr

Why do no 'AI safety' people ever say how LLM AIs failed to live up to the hype? Why do the safety people never share the failures of LLMs? …

Why do no 'AI safety' people ever say how LLM AIs failed to live up to the hype? Why do the safety people never share the failures of LLMs? This link is from Nov 2023 & shows how tech CEOs used 'safet

Why Low-Precision Transformer Training Fails: An Analysis on Flash Attention

arXiv:2510.04212v3 Announce Type: replace Abstract: The pursuit of computational efficiency has driven the adoption of low-precision formats for training transformer models. However, this progress is

ZoFia: Zero-Shot Fake News Detection with Entity-Guided Retrieval and Multi-LLM Interaction

arXiv:2511.01188v2 Announce Type: replace Abstract: The rapid spread of fake news threatens social stability and public trust, highlighting the urgent need for its effective detection. Although large

A Systematic Study of Training-Free Methods for Trustworthy Large Language Models

arXiv:2604.15789v1 Announce Type: new Abstract: As Large Language Models (LLMs) receive increasing attention and are being deployed across various domains, their potential risks, including generating

A Wasserstein Geometric Framework for Hebbian Plasticity

arXiv:2604.16052v1 Announce Type: cross Abstract: We introduce the Tan-HWG framework (Hebbian-Wasserstein-Geometry), a geometric theory of Hebbian plasticity in which memory states are modeled as prob

Adaptive multi-fidelity optimization with fast learning rates

arXiv:2604.16239v1 Announce Type: cross Abstract: In multi-fidelity optimization, biased approximations of varying costs of the target function are available. This paper studies the problem of optimiz

Aerial Multi-Functional RIS in Fluid Antennas-Aided Full-Duplex Networks: A Self-Optimized Hybrid Deep Reinforcement Learning Approach

arXiv:2604.14309v2 Announce Type: replace-cross Abstract: To address high data traffic demands of sixth-generation (6G) networks, this paper proposes a novel architecture that integrates autonomous ae

💯 agree

💯 agree The AI bubble will burst says economist Ann Pettifor She is Chair of the Political Economy Research Centre's Advisory Board at Goldsmiths, a fellow of the New Economics Foundation, a director

AI Agents and Hard Choices

arXiv:2504.15304v2 Announce Type: replace Abstract: Can AI agents deal with hard choices -- cases where options are incommensurable because multiple objectives are pursued simultaneously? Adopting a t

“AI is at best a functional mimic, not a conscious experiencing subject. …. The real moral issue lies not in making AI conscious …. but in a…

“AI is at best a functional mimic, not a conscious experiencing subject. …. The real moral issue lies not in making AI conscious …. but in avoiding transforming humans into zombies” @GaryMarcus @OEIAC

AIFIND: Artifact-Aware Interpreting Fine-Grained Alignment for Incremental Face Forgery Detection

arXiv:2604.16207v1 Announce Type: cross Abstract: As forgery types continue to emerge consistently, Incremental Face Forgery Detection (IFFD) has become a crucial paradigm. However, existing methods t

Amazon’s AI coding boom is creating a mess. Vibe coding their way to chaos.

Amazon’s AI coding boom is creating a mess. Vibe coding their way to chaos. @GaryMarcus 🦔The internal document was reported by Business Insider. Here’s the link: https://www.businessinsider.com/ai-spr

AutoDrive-R^2: Incentivizing Reasoning and Self-Reflection Capacity for VLA Model in Autonomous Driving

arXiv:2509.01944v3 Announce Type: replace-cross Abstract: Vision-Language-Action (VLA) models in autonomous driving systems have recently demonstrated transformative potential by integrating multimoda

C-Mining: Unsupervised Discovery of Seeds for Cultural Data Synthesis via Geometric Misalignment

arXiv:2604.15675v1 Announce Type: new Abstract: Achieving cultural alignment in Large Language Models (LLMs) increasingly depends on synthetic data generation. For such synthesis, the most vital initi

CASR: A Robust Cyclic Framework for Arbitrary Large-Scale Super-Resolution with Distribution Alignment and Self-Similarity Awareness

arXiv:2602.22159v2 Announce Type: replace Abstract: Arbitrary-Scale SR (ASISR) remains fundamentally limited by cross-scale distribution shift: once the inference scale leaves the training range, nois

Causal Bootstrapped Alignment for Unsupervised Video-Based Visible-Infrared Person Re-Identification

arXiv:2604.15631v1 Announce Type: new Abstract: VVI-ReID is a critical technique for all-day surveillance, where temporal information provides additional cues beyond static images. However, existing a

Cognitive Agency Surrender: Defending Epistemic Sovereignty via Scaffolded AI Friction

arXiv:2603.21735v2 Announce Type: replace-cross Abstract: The proliferation of Generative Artificial Intelligence has transformed benign cognitive offloading into a systemic risk of cognitive agency s

Comparing the latent features of universal machine-learning interatomic potentials

arXiv:2512.05717v3 Announce Type: replace-cross Abstract: The past few years have seen the development of ``universal'' machine-learning interatomic potentials (uMLIPs) capable of approximating the gr

Concept-wise Attention for Fine-grained Concept Bottleneck Models

arXiv:2604.15748v1 Announce Type: new Abstract: Recently impressive performance has been achieved in Concept Bottleneck Models (CBM) by utilizing the image-text alignment learned by a large pre-traine

Constant-Factor Approximations for Doubly Constrained Fair k-Center, k-Median and k-Means

arXiv:2604.16061v1 Announce Type: cross Abstract: We study discrete k-clustering problems in general metric spaces that are constrained by a combination of two different fairness conditions within the

Contact-Aware Planning and Control of Continuum Robots in Highly Constrained Environments

arXiv:2604.15638v1 Announce Type: new Abstract: Continuum robots are well suited for navigating confined and fragile environments, such as vascular or endoluminal anatomy, where contact with surroundi

safetyarxiv-cs-ro

Deliberative Searcher: Improving LLM Reliability via Reinforcement Learning with constraints

arXiv:2507.16727v3 Announce Type: replace Abstract: Improving the reliability of large language models (LLMs) is critical for deploying them in real-world scenarios. In this paper, we propose extbf{De

Distribution Shift Alignment Helps LLMs Simulate Survey Response Distributions

arXiv:2510.21977v2 Announce Type: replace Abstract: Large language models (LLMs) offer a promising way to simulate human survey responses, potentially reducing the cost of large-scale data collection.

Dynamic Sampling that Adapts: Self-Aware Iterative Data Persistent Optimization for Mathematical Reasoning

arXiv:2505.16176v2 Announce Type: replace Abstract: In mathematical reasoning, data selection strategies predominantly rely on static, externally defined metrics, which fail to adapt to the evolving c

DyTact: Capturing Dynamic Contacts in Hand-Object Manipulation

arXiv:2506.03103v2 Announce Type: replace Abstract: Reconstructing dynamic hand-object contacts is essential for realistic manipulation in AI character animation, XR, and robotics, yet it remains chal

Elucidating the SNR-t Bias of Diffusion Probabilistic Models

arXiv:2604.16044v1 Announce Type: new Abstract: Diffusion Probabilistic Models have demonstrated remarkable performance across a wide range of generative tasks. However, we have observed that these mo

Enhancing AI and Dynamical Subseasonal Forecasts with Probabilistic Bias Correction

arXiv:2604.16238v1 Announce Type: new Abstract: Decision-makers rely on weather forecasts to plant crops, manage wildfires, allocate water and energy, and prepare for weather extremes. Today, such for

Exploitation Over Exploration: Unmasking the Bias in Linear Bandit Recommender Offline Evaluation

arXiv:2507.18756v2 Announce Type: replace Abstract: Multi-Armed Bandit (MAB) algorithms are widely used in recommender systems that require continuous, incremental learning. A core aspect of MABs is t

Find, Fix, Reason: Context Repair for Video Reasoning

arXiv:2604.16243v1 Announce Type: new Abstract: Reinforcement learning has advanced video reasoning in large multi-modal models, yet dominant pipelines either rely on on-policy self-exploration, which

FineSteer: A Unified Framework for Fine-Grained Inference-Time Steering in Large Language Models

arXiv:2604.15488v1 Announce Type: cross Abstract: Large language models (LLMs) often exhibit undesirable behaviors, such as safety violations and hallucinations. Although inference-time steering offer

Flexible Empowerment at Reasoning with Extended Best-of-N Sampling

arXiv:2604.15614v1 Announce Type: new Abstract: This paper proposes a novel method that incorporates empowerment when reasoning actions in reinforcement learning (RL), thereby achieving the flexibilit

Follow the Flow: On Information Flow Across Textual Tokens in Text-to-Image Models

arXiv:2504.01137v3 Announce Type: replace Abstract: Text-to-image generation models suffer from alignment problems, where generated images fail to accurately capture the objects and relations in the t

From Competition to Coopetition: Coopetitive Training-Free Image Editing Based on Text Guidance

arXiv:2604.15948v1 Announce Type: new Abstract: Text-guided image editing, a pivotal task in modern multimedia content creation, has seen remarkable progress with training-free methods that eliminate

From Intention to Text: AI-Supported Goal Setting in Academic Writing

arXiv:2604.15800v1 Announce Type: cross Abstract: This study presents WriteFlow, an AI voice-based writing assistant designed to support reflective academic writing through goal-oriented interaction.

Geometric regularization of autoencoders via observed stochastic dynamics

arXiv:2604.16282v1 Announce Type: new Abstract: Stochastic dynamical systems with slow or metastable behavior evolve, on long time scales, on an unknown low-dimensional manifold in high-dimensional am

GroupDPO: Memory efficient Group-wise Direct Preference Optimization

arXiv:2604.15602v1 Announce Type: new Abstract: Preference optimization is widely used to align Large Language Models (LLMs) with preference feedback. However, most existing methods train on a single

Hierarchical Codec Diffusion for Video-to-Speech Generation

arXiv:2604.15923v1 Announce Type: cross Abstract: Video-to-Speech (VTS) generation aims to synthesize speech from a silent video without auditory signals. However, existing VTS methods disregard the h

How people use Copilot for Health

arXiv:2604.15331v1 Announce Type: cross Abstract: We analyze over 500,000 de-identified health-related conversations with Microsoft Copilot from January 2026 to characterize what people ask conversati

Import AI 454: Automating alignment research; safety study of a Chinese model; HiFloat4

This newsletter covers three main topics: advances in automating alignment research to improve AI safety processes, a safety evaluation study of a Chinese AI model, and technical details about HiFloat

safetyimport-ai

Importantly, the tweet by @itsolelehmann may be misleading. I stand by what I said but have some reason to think that @AmandaAskell’s views …

Importantly, the tweet by @itsolelehmann may be misleading. I stand by what I said but have some reason to think that @AmandaAskell’s views may have been misrepresented. Seeking clarification and will

Improving Reasoning Capabilities in Small Models through Mixture-of-Layers Distillation with Stepwise Attention on Key Information

arXiv:2604.15701v1 Announce Type: new Abstract: The significant computational demands of large language models have increased interest in distilling reasoning abilities into smaller models via Chain-o

Information-Consistent Language Model Recommendations through Group Relative Policy Optimization

arXiv:2512.12858v3 Announce Type: replace-cross Abstract: Large Language Models (LLMs) are increasingly deployed in business-critical domains such as finance, education, healthcare, and customer suppo

Jailbreak Scaling Laws for Large Language Models: Polynomial-Exponential Crossover

arXiv:2603.11331v2 Announce Type: replace-cross Abstract: Adversarial attacks can reliably steer safety-aligned large language models toward unsafe behavior. Empirically, we find that strong adversari