safety

The Wittgensteinian Representation Hypothesis: Is Language the Attractor of Multimodal Convergence?

arXiv:2605.09352v1 Announce Type: new Abstract: Understanding why independently trained neural networks from different modalities converge toward shared representations, and where this convergence lea

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

There will be no AI jobpocalypse. The story that AI will lead to massive unemployment is stoking unnecessary fear. AI — like any other techn…

DGX agent

There will be no AI jobpocalypse. The story that AI will lead to massive unemployment is stoking unnecessary fear. AI — like any other technology — does affect jobs, but telling overblown stories of l

safetyandrew-ng--x

12 May 2026

Thermal-Det: Language-Guided Cross-Modal Distillation for Open-Vocabulary Thermal Object Detection

arXiv:2605.10130v1 Announce Type: new Abstract: Existing open-vocabulary detectors focus on RGB images and fail to generalize to thermal imagery, where low texture and emissivity variations challenge

TIE: Time Interval Encoding for Video Generation over Events

arXiv:2605.10543v1 Announce Type: new Abstract: Director-style prompting, robotic action prediction, and interactive video agents demand temporal grounding over concurrent events -- a regime in which

tl;dr - let's not just remove negative behavior from models, but also add positive ones 👍

tl;dr - let's not just remove negative behavior from models, but also add positive ones 👍 If anyone builds it, everyone thrives. Over the past decade, a lot of important work on AI alignment has focus

safetyyohei-nakajima--x

TodyComm: Task-Oriented Dynamic Communication for Multi-Round LLM-based Multi-Agent System

arXiv:2602.03688v2 Announce Type: replace Abstract: Multi-round LLM-based multi-agent systems rely on effective communication structures to support collaboration across rounds. However, most existing

Token Buncher: Shielding LLMs from Harmful Reinforcement Learning Fine-Tuning

arXiv:2508.20697v3 Announce Type: replace-cross Abstract: As large language models (LLMs) continue to grow in capability, so do the risks of harmful misuse through fine-tuning. While most prior studie

Total Generalized Variation regularization closes the gap between neural-eld and classical methods in seismic travel-time tomography

arXiv:2605.09960v1 Announce Type: cross Abstract: Travel-time tomography forces a trade-off between mesh resolution and stability in which the regularizer choice dominates what can be recovered. We in

Toward Reliable Sim-to-Real Predictability for MoE-based Robust Quadrupedal Locomotion

arXiv:2602.00678v4 Announce Type: replace Abstract: Reinforcement learning has shown strong promise for quadrupedal agile locomotion, even with proprioception-only sensing. In practice, however, sim-t

Towards Customized Multimodal Role-Play

arXiv:2605.08129v1 Announce Type: new Abstract: Unified multimodal understanding and generation models enable richer human-AI interaction. Yet jointly customizing a character's persona, dialogue style

TRACE: Distilling Where It Matters via Token-Routed Self On-Policy Alignment

arXiv:2605.10194v1 Announce Type: new Abstract: On-policy self-distillation (self-OPD) densifies reinforcement learning with verifiable rewards (RLVR) by letting a policy teach itself under privileged

Training-Free Cultural Alignment of Large Language Models via Persona Disagreement

arXiv:2605.10843v1 Announce Type: cross Abstract: Large language models increasingly mediate decisions that turn on moral judgement, yet a growing body of evidence shows that their implicit preference

Training with Harnesses: On-Policy Harness Self-Distillation for Complex Reasoning

arXiv:2605.08741v1 Announce Type: new Abstract: Inference-time harnesses substantially improve large language models on complex reasoning tasks. However, the intrinsic capabilities of the underlying m

Trajectory-Consistent Flow Matching for Robust Visuomotor Policy Learning

arXiv:2605.08511v1 Announce Type: new Abstract: Flow matching policies learn continuous velocity fields that transport noise to actions, enabling fast deterministic inference for robot manipulation. H

TrajTok: Learning Trajectory Tokens enables better Video Understanding

arXiv:2602.22779v2 Announce Type: replace Abstract: Tokenization in video models, typically through patchification, generates an excessive and redundant number of tokens. This severely limits video ef

TripleWin: Fixed-Point Equilibrium Pricing for Data-Model Coupled Markets

arXiv:2511.03368v2 Announce Type: replace Abstract: The rise of the machine learning (ML) model economy has intertwined markets for training datasets and pre-trained models. However, most pricing appr

Trustworthy AI: Ensuring Reliability and Accountability from Models to Agents

arXiv:2605.08964v1 Announce Type: new Abstract: In this thesis, we develop algorithms with theoretical guarantees for ensuring reliability and accountability of Machine Learning (ML) systems. As ML sy

UAV-Assisted Scan-to-Simulation for Landslides Using Physics-Informed Gaussian Splatting

arXiv:2605.10715v1 Announce Type: new Abstract: Landslide monitoring and simulation play an important role in urban safety assessment and disaster prevention. Existing landslide simulation pipelines t

Uncertainty-Aware and Decoder-Aligned Learning for Video Summarization

arXiv:2605.09507v1 Announce Type: new Abstract: Video summarization aims to produce a compact representation of a long video by selecting a subset of temporally important segments that best reflect hu

Uni-Hand: Universal Hand Motion Forecasting in Egocentric Views

arXiv:2511.12878v4 Announce Type: replace Abstract: Forecasting how human hands move in egocentric views is critical for applications like augmented reality and human-robot policy transfer. Recently,

Unified Noise Steering for Efficient Human-Guided VLA Adaptation

arXiv:2605.10821v1 Announce Type: new Abstract: Diffusion-based vision-language-action (VLA) models have emerged as strong priors for robotic manipulation, yet adapting them to real-world distribution

Uniform Inductive Spatio-Temporal Kriging

arXiv:2603.05301v2 Announce Type: replace Abstract: Inductive spatio-temporal kriging infers signals at unobserved locations from observed sensors, but real-world observations are often incomplete and

Unison: Harmonizing Motion, Speech, and Sound for Human-Centric Audio-Video Generation

arXiv:2605.08729v1 Announce Type: new Abstract: Motion, speech, and sound effects are fundamental elements of human-centric videos, yet their heterogeneous temporal characteristics make joint generati

Unlearners Can Lie: Evaluating and Improving Honesty in LLM Unlearning

arXiv:2605.08765v1 Announce Type: cross Abstract: Unlearning in large language models (LLMs) aims to remove harmful training data while preserving overall utility. However, we find that existing metho

Unsupervised Process Reward Models

arXiv:2605.10158v1 Announce Type: new Abstract: Process Reward Models (PRMs) are a powerful mechanism for steering large language model reasoning by providing fine-grained, step-level supervision. How

Upholding Epistemic Agency: A Brouwerian Assertibility Constraint for Responsible AI

arXiv:2603.03971v2 Announce Type: replace-cross Abstract: Generative AI can convert uncertainty into hypersuasive, authoritative-seeming verdicts, displacing the justificatory work on which democratic

Users as Annotators: LLM Preference Learning from Comparison Mode

arXiv:2510.13830v2 Announce Type: replace-cross Abstract: Pairwise preference data have played an important role in the alignment of large language models (LLMs). Each sample of such data consists of

V-ABS: Action-Observer Driven Beam Search for Dynamic Visual Reasoning

arXiv:2605.10172v1 Announce Type: cross Abstract: Multimodal large language models (MLLMs) have achieved remarkable success in general perception, yet complex multi-step visual reasoning remains a per

Value-Decomposed Reinforcement Learning Framework for Taxiway Routing with Hierarchical Conflict-Aware Observations

arXiv:2605.08754v1 Announce Type: new Abstract: Taxiway routing and on-surface conflict avoidance are coupled safety-critical decision problems in airport surface operations. Existing planning and opt

Variational Inference for Levy Process-Driven SDEs via Neural Tilting

arXiv:2605.10934v1 Announce Type: cross Abstract: Modelling extreme events and heavy-tailed phenomena is central to building reliable predictive systems in domains such as finance, climate science, an

Verification Mirage: Mapping the Reliability Boundary of Self-Verification in Medical VQA

arXiv:2605.10850v1 Announce Type: new Abstract: Self-verification, re-invoking the same vision language model (VLM) in a fresh context to check its own generated answer, is increasingly used as a defa

Verifier-Free RL for LLMs via Intrinsic Gradient-Norm Reward

arXiv:2605.09920v1 Announce Type: cross Abstract: While Reinforcement Learning with Verifiable Rewards (RLVR) has recently emerged as a promising post-training paradigm for Large Language Models (LLMs

VISTA: A Generative Egocentric Video Framework for Daily Assistance

arXiv:2605.10579v1 Announce Type: new Abstract: Training AI agents to proactively assist humans in daily activities, from routine household tasks to urgent safety situations, requires large-scale visu

Wavelet Policy: Imitation Learning in the Scale Domain with World Prior Memory

arXiv:2504.04991v4 Announce Type: replace Abstract: Conventional visuomotor imitation learning usually predicts future robot actions directly in the time domain. Such formulations often have limited p

'We're going to reach a point of diminishing returns from scaling. That's what actually happened. The industry knows this. But doesn't want …

'We're going to reach a point of diminishing returns from scaling. That's what actually happened. The industry knows this. But doesn't want to tell you.' @GaryMarcus, AI Expert, Scientist & Author at

safetygary-marcus--x

What should post-training optimize? A test-time scaling law perspective

arXiv:2605.10716v1 Announce Type: new Abstract: Large language models are increasingly deployed with test-time strategies: sample N responses, score them with a reward model or verifier, and return th

What Structural Inductive Bias Helps Transformers Reason Over Knowledge Graphs? A Study with Tabula RASA

arXiv:2602.02834v3 Announce Type: replace-cross Abstract: What structural inductive bias helps transformers reason over knowledge graphs? Through controlled ablations of a minimal transformer modifica

When a Robot is More Capable than a Human: Learning from Constrained Demonstrators

arXiv:2510.09096v3 Announce Type: replace-cross Abstract: Learning from demonstrations enables experts to teach robots complex tasks using interfaces such as kinesthetic teaching, joystick control, an

When Agents Overtrust Environmental Evidence: An Extensible Agentic Framework for Benchmarking Evidence-Grounding Defects in LLM Agents

arXiv:2605.08828v1 Announce Type: new Abstract: Large language model agents increasingly operate through environment-facing scaffolds that expose files, web pages, APIs, and logs. These observations i

When Can Digital Personas Reliably Approximate Human Survey Findings?

arXiv:2605.10659v1 Announce Type: cross Abstract: Digital personas powered by Large Language Models (LLMs) are increasingly proposed as substitutes for human survey respondents, yet it remains unclear

When Language Overwrites Vision: Over-Alignment and Geometric Debiasing in Vision-Language Models

arXiv:2605.08245v1 Announce Type: cross Abstract: Vision-Language Models (VLMs) increasingly power high-stakes applications, from medical imaging to autonomous systems, yet they routinely hallucinate,

When More Parameters Hurt: Foundation Model Priors Amplify Worst-Client Disparity Under Extreme Federated Heterogeneity

arXiv:2605.08992v1 Announce Type: new Abstract: Federated learning (FL) is increasingly used to fine-tune foundation models (FMs) on distributed private data. The community largely assumes that large-

Where Do Flow Semantics Reside? A Protocol-Native Tabular Pretraining Paradigm for Encrypted Traffic Classification

arXiv:2603.10051v2 Announce Type: replace-cross Abstract: Self-supervised masked modeling shows promise for encrypted traffic classification by masking and reconstructing raw bytes. Yet recent work re

White Circle raises $11M to help companies secure and monitor AI model behavior

Artificial intelligence guardrail and monitoring startup Pumpkin Intelligence Inc., which operates as White Circle, announced today it raised 11 million in seed funding from a who’s who of AI leadersh

safetysiliconangle

Why Adam Works Better with eta_1 = eta_2: The Missing Gradient Scale Invariance Principle

arXiv:2601.21739v2 Announce Type: replace-cross Abstract: Adam has been at the core of large-scale training for almost a decade, yet a simple empirical fact remains unaccounted for: both validation sc

Why Do DiT Editors Drift? Plug-and-Play Low Frequency Alignment in VAE Latent Space

arXiv:2605.08250v1 Announce Type: cross Abstract: Recent advances in diffusion transformers (DiTs) have enabled promising single-turn image editing capabilities. However, multi-turn editing often lead

WISTERIA: Learning Clinical Representations from Noisy Supervision via Multi-View Consistency in Electronic Health Records

arXiv:2605.09765v1 Announce Type: cross Abstract: Representation learning in electronic health records (EHR) has largely followed paradigms inherited from natural language processing, relying on seque

wow! “It is the [House] Committee’s understanding that the new board of directors at OpenAI tried to address these problems upon your return…

wow! “It is the [House] Committee’s understanding that the new board of directors at OpenAI tried to address these problems upon your return by creating an “audit committee to review potential conflic

safetygary-marcus--x