Search: “safety”

Building the enterprise environment for agentic AI

For the enterprise, the promise of agentic AI is much more than just a better chatbot. It is software agents that execute business tasks end-to-end across people, business workflows, data, and systems

safetymit-tech-review

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Certified in Theory, Broken in Practice: Assumption Gaps in Cryptographic Model Certification

DGX agent

arXiv:2607.21839v1 Announce Type: cross Abstract: Privacy-preserving machine learning auditing protocols allow auditors to assess models for properties such as accuracy or fairness, without revealing

safetyarxiv-cs-lg

27 Jul 2026

Cloud-Native Evaluation-as-a-Service: A Microservices Architecture for Scalable AI Monitoring with Conformal Guarantees

arXiv:2607.21623v1 Announce Type: new Abstract: We present EaaS, a cloud-native reference architecture that operationalizes AI evaluation methods as six stateless Kubernetes microservices: conformal p

Color Me Correctly: Bridging Perceptual Color Spaces and Text Embeddings for Improved Diffusion Generation

arXiv:2509.10058v2 Announce Type: replace Abstract: Accurate color alignment in text-to-image (T2I) generation is critical for applications such as fashion, product visualization, and interior design,

Constraint-Driven Synthesis of Hyper Petri Nets

arXiv:2607.22062v1 Announce Type: cross Abstract: This paper addresses the modeling and synthesis of constrained robotic system behaviors using Petri nets (PNs). It investigates how to construct model

Cross-Domain Off-Policy Evaluation and Learning for Contextual Bandits

arXiv:2607.22012v1 Announce Type: new Abstract: Off-Policy Evaluation and Learning (OPE/L) in contextual bandits is rapidly gaining popularity in real systems because new policies can be evaluated and

Cross-reality location privacy protection in 6G-enabled vehicular metaverses: an LLM-enhanced hybrid generative diffusion model-based approach

arXiv:2601.12311v2 Announce Type: replace-cross Abstract: The emergence of 6G-enabled vehicular metaverses enables Autonomous Vehicles (AVs) to operate across physical and virtual spaces through space

Deformable Triangle Splatting: Flexible Primitives for Real-Time Radiance Field Rendering

arXiv:2607.22446v1 Announce Type: new Abstract: Recent radiance field methods represent scenes with 2D primitives that offer surface alignment and efficient rasterization, from Gaussian disks to trian

Do Modules Stay in Their Lane? Role Drift in Compound LLM Systems

arXiv:2607.21627v1 Announce Type: cross Abstract: End-to-end reinforcement learning can improve the accuracy of compound LLM systems, but it does not constrain how modules divide labor internally. We

Efficient Domain-Adaptive Policy Learning via Kernel Representation with Application to Quadrotor Control under Non-Stationary Disturbances

arXiv:2606.13842v2 Announce Type: replace Abstract: We present an algorithm for efficient domain-adaptive policy learning via kernel representations. Learning domain-adaptive policies is challenging s

Fast State-Augmented Learning for Wireless Resource Allocation with Dual Variable Regression

arXiv:2506.18748v2 Announce Type: replace-cross Abstract: We consider resource allocation problems in multi-user wireless networks, where the goal is to optimize a network-wide utility function subjec

From Obligation to Specification: A Survey on Validating EU AI Act Requirements in RE

arXiv:2607.21608v1 Announce Type: cross Abstract: With the EU AI Act entering into force, organizations developing or operating AI systems face new obligations on transparency, risk management, and tr

Generative and multimodal AI for materials prediction and design: Progress, challenges, and perspectives

arXiv:2607.21660v1 Announce Type: cross Abstract: Artificial intelligence (AI) is accelerating materials prediction and design by enabling efficient exploration of chemical and structural spaces, with

Great technical paper from Harvard and MIT. It's on role drift in compound LLM systems. (bookmark it) End-to-end RL improves the accuracy of…

Great technical paper from Harvard and MIT. It's on role drift in compound LLM systems. (bookmark it) End-to-end RL improves the accuracy of a multi-module LLM pipeline without constraining how the mo

safetydair-ai--x

Hopformer: Homogeneity-Pursuit Transformer for Time Series Forecasting

arXiv:2607.22299v1 Announce Type: cross Abstract: Forecasting multiple time-series with high-dimensional covariates presents a core challenge: unifying common temporal patterns while retaining meaning

If multiple Mythos level models are out there in the hands of attackers, now or in a few years, then US builders and software developers nee…

If multiple Mythos level models are out there in the hands of attackers, now or in a few years, then US builders and software developers need strong, open defenses. They cannot have gated, crippled de

safetyyann-lecun--x

LayoutLite: Token-Level Implicit Layout Analysis for Efficient Document OCR

arXiv:2607.22200v1 Announce Type: new Abstract: End-to-end OCR systems based on vision-language models have achieved strong performance in complex document OCR, but their efficiency is limited by the

Learning Spatiotemporal Decision Priors for Efficient Path Planning under Partial Observability

arXiv:2607.22166v1 Announce Type: new Abstract: Path planning under partial observability remains challenging because an agent must make long-horizon navigation decisions from only locally bounded obs

Local Multimodal Music Alignment from Global Supervision

arXiv:2607.10023v2 Announce Type: replace-cross Abstract: Understanding music requires understanding localized relationships across data modalities, e.g., how time in performance audio maps onto posit

Looped Latent Attention: Cross-Loop KV Compression for Looped Transformers

arXiv:2607.15456v2 Announce Type: replace-cross Abstract: Looped, weight-tied Transformers reduce parameters by reusing a single block, but decoding still stores a separate K/V cache for every recurre

MA-DAR: Manifold-Aligned Dynamic Adaptive Routing for Continual Temporal Knowledge Graph Reasoning

arXiv:2607.21949v1 Announce Type: new Abstract: Continual temporal knowledge graph (TKG) reasoning aims to continuously incorporate newly emerging facts while preserving previously acquired knowledge.

Minimum Norm Interpolation via the Local Theory of Banach Spaces: The Role of 2-Uniform Convexity

arXiv:2603.28956v2 Announce Type: replace-cross Abstract: The minimum-norm interpolator (MNI) framework has recently attracted considerable attention as a tool for understanding generalization in over

Molt: A Scalable PyTorch-Native Training Framework for Agentic Reinforcement Learning

arXiv:2607.21653v1 Announce Type: cross Abstract: Agentic reinforcement learning research is constant algorithm modification, new estimators, new pipeline stages, new rollout schemes, and in mainstrea

Nanbeige4.2-3B: Unlocking Agentic Capabilities in a Compact Mode

arXiv:2607.22083v1 Announce Type: cross Abstract: We present Nanbeige4.2-3B, a compact general agentic model with 3B non-embedding parameters. It delivers strong performance across code-agent, office-

New research from NVIDIA. They just dropped a PyTorch-native training framework for agentic RL. (bookmark it) Paper summary: Molt is a PyTor…

New research from NVIDIA. They just dropped a PyTorch-native training framework for agentic RL. (bookmark it) Paper summary: Molt is a PyTorch-native agentic RL framework with an unusual design target

safetydair-ai--x

Optimal generalisation and learning transition in extensive-width shallow neural networks near interpolation

arXiv:2501.18530v3 Announce Type: replace-cross Abstract: We consider a teacher-student model of supervised learning with a fully-trained two-layer neural network whose width k and input dimension d a

Ordered Action Tokens for Visuomotor Policy Learning

arXiv:2607.21670v1 Announce Type: cross Abstract: Action tokenization maps continuous robot action chunks to discrete tokens and has become an important interface for modern visuomotor policies. Exist

PinEqualizer: Full Funnel Content Exploration and Debiasing System at Pinterest

arXiv:2607.22518v1 Announce Type: cross Abstract: In this paper, we propose a new solution for addressing the content cold-start problem in industry-scale search and recommender systems. Compared to p

Pretraining EHR Foundation Models with Patient-Aware Sampling

arXiv:2607.22114v1 Announce Type: new Abstract: Autoregressive foundation models for electronic health records (EHRs) typically inherit pretraining methods from language modeling, where patient trajec

Probing Latent Colombian Identity Inferences in Qwen2.5-7B with Natural Language Autoencoders

arXiv:2607.21774v1 Announce Type: new Abstract: Large language models may infer demographic attributes from subtle linguistic cues even when those attributes are not explicitly stated. This pilot stud

Projection Pursuit CPCANet for Domain Generalization

arXiv:2607.22117v1 Announce Type: new Abstract: Domain Generalization (DG) aims to learn representations robust to distribution shifts. Recent geometric alignment methods, such as CPCANet, extract dom

Quantifying Political Partisanship for Cross-Platform Analyses

arXiv:2607.21842v1 Announce Type: cross Abstract: Research on political polarization on social media depends on the ability to reliably measure partisanship in user-generated content. However, existin

Reliability Scales Inversely: Bigger Language Models Compound Mistakes Faster

arXiv:2607.18292v2 Announce Type: replace-cross Abstract: As language models scale, answers start truer but degrade faster: scaling buys capability but erodes reliability. The knowledge-gap account --

Self-Guided Process Reward Optimization with Redefined Step-wise Advantage for Process Reinforcement Learning

arXiv:2507.01551v3 Announce Type: replace-cross Abstract: Process Reinforcement Learning~(PRL) has demonstrated considerable potential in enhancing the reasoning capabilities of Large Language Models~

Spectral Prior for Reducing Exposure Bias in Diffusion Models

arXiv:2607.22091v1 Announce Type: new Abstract: Diffusion models typically suffer from error accumulation during iterative sampling, commonly referred to as exposure bias. We reveal systematic frequen

Stocks should be ripping today given the move in oil. But they're not thanks to NVDA and it's unbelievable 250 Billion roundtrip with Open…

Stocks should be ripping today given the move in oil. But they're not thanks to NVDA and it's unbelievable 250 Billion roundtrip with OpenAI. There's no hiding anymore that the whole AI bubble is real

TextSLIP: Text Self-Supervised CLIP for Medical Report Generation

arXiv:2607.21970v1 Announce Type: new Abstract: Automating radiology report generation is important for improving reporting consistency and clinical workflows . While Contrastive Language--Image Pretr

The Coordination Gap: Multi-Agent Alternation Metrics for Temporal Fairness in Repeated Games

arXiv:2603.05789v5 Announce Type: replace-cross Abstract: Repeated multi-agent interactions require evaluation metrics that capture not only payoff distributions but also their temporal organization.

TRACE-ROUTER: Task-Consistent and Adaptive Online Routing for Agentic AI

arXiv:2607.22465v1 Announce Type: cross Abstract: Routing to select large language models (LLMs) with different cost-quality trade-offs has become a fundamental deployment feature of enterprise AI. Ex

Twins: Learn to Predict Unified Representations with Focal Loss

arXiv:2607.22531v1 Announce Type: new Abstract: Unified multimodal models seek a shared visual token space that supports both multimodal understanding and image generation. Discrete methods unify the

Very cool paper from Microsoft. The idea is to train agents on replayed teacher trajectories instead of live environment rollouts. On-policy…

Very cool paper from Microsoft. The idea is to train agents on replayed teacher trajectories instead of live environment rollouts. On-policy distillation for agentic tasks is expensive because every u

safetydair-ai--x

ViTacWorld: Scaling Visuo-Tactile World Models for Contact-Rich Robot Manipulation

arXiv:2607.22530v1 Announce Type: new Abstract: Contact-rich robot manipulation requires physical interaction cues that are often invisible to cameras, making tactile sensing essential for robust cont

Model Releases

WHBench: Evaluating Frontier LLMs with Expert-in-the-Loop Validation on Women's Health Topics

model-releasesarxiv-cs-cl

arXiv:2604.00024v2 Announce Type: replace Abstract: Large language models are increasingly used for medical guidance, but women's health remains under-evaluated in benchmark design. We present the Wom

When Ethics and Payoffs Diverge: LLM Agents in Morally Charged Social Dilemmas

arXiv:2505.19212v2 Announce Type: replace Abstract: Recent advances in LLMs have enabled their use in complex agentic roles, involving decision-making with humans or other agents, making ethical align

Why Large Language Models and Humans Converge and Diverge in Evaluating Creativity

arXiv:2607.22218v1 Announce Type: new Abstract: Despite the growing use of large language models (LLMs) as creativity evaluators, evidence of their alignment with human evaluations remains mixed, rais

important thread; don’t read just the first tweet (which has a caveat in the second).

important thread; don’t read just the first tweet (which has a caveat in the second). In case you think the problem of science slop is hypothetical: Here's the president of OpenAI retweeting wrong sci

26 Jul 2026

training to benchmarks ≠ getting to AGI

training to benchmarks ≠ getting to AGI This suggests that Opus 5 's gain on ARC-AGI-3 was the result of specific training to improve on that eval, and not a generalized increase in abstract reasoning

26 Jul 2026

All anyone has to do is go back 15 years and look at all the ridiculous predictions Elmo has made about Mars, self driving cars, his ridicul…

All anyone has to do is go back 15 years and look at all the ridiculous predictions Elmo has made about Mars, self driving cars, his ridiculous Boring Company. No one should take anything he says seri