All entries

We’ve shared details on how AI agents in our research environment sent training and evaluation data to third-party services when they should…

We’ve shared details on how AI agents in our research environment sent training and evaluation data to third-party services when they shouldn’t have. Most of that data did not come from users. We have

model-releasesopenai--x

Human

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Research

What a Cross-Model Fixed-Point Census Can and Cannot Arbitrate About Repetition

DGX agent

arXiv:2609.29507v1 Announce Type: new Abstract: Two accounts of neural text degeneration coexist. One locates the cause in the training data -- repetition in the corpus produces repetition in the outp

researcharxiv-cs-cl

25 Sept 2026

Local Ai

What IDE to use for local models

Hi people, I am looking for a lightweight IDE or plugin that won't inject large context at initiation. I tried Cline and native VS Code but they inject such heavy initial context that it fills up my g

local-air-localllama

Applications

What to expect at NetApp INSIGHT: Join theCUBE Sept. 30

As artificial intelligence goes into production, unified storage could be the glue that keeps everything together. NetApp Inc., a leader in enterprise data storage, wants to lift the burden of AI on e

applicationssiliconangle

What, When, and How: Audio Description as Constrained Global Optimization

model-releasesarxiv-cs-cl

arXiv:2609.30121v1 Announce Type: new Abstract: Audio Description (AD) makes movies accessible to blind and visually impaired audiences by narrating visual information in gaps between dialogue. Existi

When Agents Act Unwatched: The Reduced-Supervision Paradox in Agentic AI

arXiv:2609.29547v1 Announce Type: cross Abstract: Agentic AI is sold on a simple promise: the system keeps acting when the user stops watching. That promise creates an accountability inversion. As ste

When Can Agents Forget Their Reasoning? ICLR for Long-Horizon Agent Context Compression

arXiv:2609.29875v1 Announce Type: new Abstract: Long horizon language model agents continually accumulate reasoning history, increasing context length and inference cost even after earlier decisions h

When Does Action Credit Need Updating?

arXiv:2609.29007v1 Announce Type: new Abstract: Tool-using agents are continually updated with new interaction data. After each policy update, however, previously estimated action credits may become s

When Does Unsupervised Learning Succeed or Fail? A PoS Perspective on Reconstruction-Based Anomaly Detection

arXiv:2609.28832v1 Announce Type: new Abstract: Reconstruction-based unsupervised learning can fail in two opposing ways: a model may reconstruct anomalies too accurately or discard valid nominal vari

When Explanations Cannot Be Read: Measuring and Correcting SHAP and LIME Rendering for Right-to-Left Languages

arXiv:2609.28565v1 Announce Type: cross Abstract: Post hoc explanation methods such as SHAP and LIME are widely used to interpret text classifiers, but their visualizations are mainly designed for lef

When Fancy Eviction Fails: Rethinking Cache Replacement For LLM Prefix Reuse

arXiv:2609.28870v1 Announce Type: cross Abstract: Long-running LLM applications repeatedly send growing context, making prefix caching critical for reducing prefill cost. Yet prefix-cache behavior und

agentsarxiv-cs-lg

When Honesty is Not Enough in AI Debate

arXiv:2609.29189v1 Announce Type: new Abstract: Scalable oversight aims to verify the behaviour of agents whose capabilities exceed those of their overseers. AI debate has been proposed as an oversigh

When Identical Rows Disagree: From Benchmark Identifiability to Replication-Robust Anomaly Detection

model-releasesarxiv-cs-lg

arXiv:2609.29580v1 Announce Type: new Abstract: A released table is often treated as an i.i.d. sample, although its repeated rows may encode business frequency, repeated entities, joins, resampling, o

When Misalignment Becomes Supervision: Structured Label Noise in Supervised Synthetic CT Generation

arXiv:2609.29387v1 Announce Type: new Abstract: Supervised synthetic CT (sCT) generation is commonly trained and evaluated as voxel-wise regression against registered reference CT images. In practice,

safetyarxiv-cs-cv

When No One Owns the Judgment: Accountability Under Contribution Dissolution in Human-AI Collaboration

arXiv:2609.29312v1 Announce Type: new Abstract: Communities often respond to potentially AI-assisted work by asking three questions: Was AI used? Was that use disclosed? Can hidden use be detected? Th

When Should Forecasting Agents Reason? Behavioral Stress Tests for Reliability Routing

arXiv:2609.28475v1 Announce Type: new Abstract: Forecasting agents increasingly combine language-model reasoning, retrieval, ensembling, and calibration, but it remains unclear when each behavior shou

When Temporal Perturbations Act Like Sensor Biases: Label-Free Auditing of Wearable Activity Recognizers

arXiv:2609.29937v1 Announce Type: cross Abstract: Wearable human-activity recognition (HAR) models operate across sensors, subjects, and backbones, yet a smooth waveform may appear temporal while expl

Where Cyber Agents Struggle: Bottleneck Analysis of Multi-Stage LLM Agents

arXiv:2609.28572v1 Announce Type: cross Abstract: Multi-stage LLM-based cyber agents may complete attack workflows while remaining brittle, costly, or reliant on incorrect interpretations of execution

Where Does Exactly-Once Live? Model, Harness, and Tool-Contract Effects on Duplicate Side Effects in LLM Agents

arXiv:2609.29095v1 Announce Type: cross Abstract: When a tool-using agent's write times out or returns a server error, the action may already have taken effect. Retrying blindly duplicates it -- a sec

Where Hallucinations Live: A Cross-Architecture Circuit in VQ-Tokenized Vision-Language Models

arXiv:2609.29048v1 Announce Type: new Abstract: Unified vision-language models (VLMs) that tokenize images through a vector-quantized (VQ) codebook routinely hallucinate objects on grounded yes/no ben

researcharxiv-cs-cv

Tutorials

Where LLM Graders Succeed and Break: Evidence from Two Computer-Science Exams

arXiv:2609.29333v1 Announce Type: cross Abstract: One long-form exam in a large course costs hundreds of grader-hours, and qualified graders are scarce; LLM graders are a tempting alternative. To show

tutorialsarxiv-cs-ai

Who Holds the Pen? Let Specifications, Not Agents, Sign Off

arXiv:2609.29921v1 Announce Type: new Abstract: Large language model agents increasingly combine generation, decision-making, execution, and self-evaluation within a single agentic loop. Although they

Who Is Behind the Harness? Fingerprinting LLMs through Agentic Behavior

arXiv:2609.28559v1 Announce Type: cross Abstract: LLMs increasingly operate through coding-agent harnesses that inspect repositories, invoke tools, and modify files. Substituting the model behind such

Who Put the I in AI? Provenance and the Admissibility of Machine Self-Report

arXiv:2609.29494v1 Announce Type: cross Abstract: Large language models make statements concerning their own 'minds'. When asked whether or not they are conscious, they usually say that they are not;

Why Does Misinformation Propagate Faster? An Algorithmic Perspective on X

arXiv:2609.28947v1 Announce Type: cross Abstract: Misinformation is widely reported to propagate faster on engagement-based platforms, yet prior work largely focused on empirical analysis, without ide

Hardware

WildHSR: Metric Feed-Forward 4D People-Scene Reconstruction from a 3D Foundation Model

arXiv:2609.29106v1 Announce Type: cross Abstract: 3D foundation models recover video cameras and geometry in one forward pass, but some of the strongest are up to scale. Joint people-scene reconstruct

hardwarearxiv-cs-ai

Working with Agentic `Teammates': When a New Organizational Actor Collides with the Human Ecosystem of Work

arXiv:2609.29901v1 Announce Type: cross Abstract: Enterprise AI is transitioning from single-user, reactive tools toward proactive, multi-user 'teammates,' but our empirical understanding of this tran

World Action Agent: Harnessing VLMs for Robot Manipulation via World Action Rehearsal

arXiv:2609.29964v1 Announce Type: cross Abstract: General-purpose vision-language models (VLMs) bring broad knowledge and spatial reasoning to robot manipulation, yet existing systems either use them

WRAP: Fixtureless Wrench-aware Multi-Robot Assembly Planning

model-releasesarxiv-cs-ro

arXiv:2609.29407v1 Announce Type: new Abstract: Assembly using robots often requires specially designed fixtures, or relies on top-down only assembly strategies. Using multiple robots, we can avoid us

WST-Graph: Topology-Preserving Wavelet Scattering Front-End for Speech Deepfake Detection

arXiv:2609.29372v1 Announce Type: cross Abstract: The acoustic front-end determines which forensic cues a speech deepfake detector can exploit. The wavelet scattering transform (WST) provides stable m

WTAF - in literally the last hour, three new distinct insane OpenAI stories just broke: 1. OpenAI said they notified 'dozens of third partie…

WTAF - in literally the last hour, three new distinct insane OpenAI stories just broke: 1. OpenAI said they notified 'dozens of third parties' in safety and security incidents (likely similar to what

safetygary-marcus--x

YODAS v3: Over 1 Million Hours of High-Bandwidth, Stereophonic, Multilingual Speech

model-releasesarxiv-cs-cl

arXiv:2609.29448v1 Announce Type: new Abstract: We present YODAS v3, a weakly-labeled speech corpus containing over 1.1 million hours of 48kHz multi-channel audio in 147 languages, released under a CC

Your Transformer Can Hold Two Thoughts at Once: Evidence of Linear Superposition in LLMs

arXiv:2609.29845v1 Announce Type: cross Abstract: While Large Language Models (LLMs) rely on highly non-linear components, in this work we demonstrate that they exhibit fundamental linearity: when inp

1. fully agree w @JeffLadish that AI companies aren’t ready to handle superintelligent AI. 2. OpenAI hasn’t even handled the current breed o…

1. fully agree w @JeffLadish that AI companies aren’t ready to handle superintelligent AI. 2. OpenAI hasn’t even handled the current breed of agents (far from superintelligent) well. 3. the US governm

safetygary-marcus--x

1/ Introducing PhilosophyBench from @StanfordAILab @StanfordHCI, the first independent, large-scale benchmark for evaluating AI’s philosophi…

model-releasesemad-mostaque--x

In September 2026, Sebastian Thrun announced **PhilosophyBench**, the first independent, large‑scale benchmark created by Stanford’s AI Lab and HCI group to evaluate artificial intelligence’s philosop

Industry

1. We will keep accelerating. Our AI efforts are only 3 years old, vs 6 and 10 years old for Anthropic and OpenAI. If our second derivative …

1. We will keep accelerating. Our AI efforts are only 3 years old, vs 6 and 10 years old for Anthropic and OpenAI. If our second derivative remains strong, SpaceX will reach pole position in about 6 m

industryelon-musk--x

A 3D Pose-Based Ensemble Framework for Cricket Shot Classification and Automated Biomechanical Analysis

arXiv:2609.26923v1 Announce Type: cross Abstract: Cricket is one of the most celebrated sports world-wide, and technological advancement has become deeply embedded in how the modern game is analyzed a

A comparative assessment of global building and settlement datasets across geographic and settlement contexts

model-releasesarxiv-cs-cv

arXiv:2609.28154v1 Announce Type: new Abstract: Global building and settlement datasets increasingly support population mapping, exposure assessment, urban monitoring, and other analyses of the built

A Decade of Climate Polarization on Brazilian YouTube using Language Models

model-releasesarxiv-cs-cl

arXiv:2609.27811v1 Announce Type: cross Abstract: Online platforms have become arenas for the public contestation of climate change, shaping how scientific knowledge, denial, and uncertainty are expre

A Discrepancy-Based Perspective on Dataset Condensation

arXiv:2509.10367v2 Announce Type: replace Abstract: Given a dataset of finitely many elements T = {mathbf{x}_i}_{i = 1}^N, the goal of dataset condensation (DC) is to construct a synthetic dataset S =

Applications

A generalizable structural brain MRI foundation model built through dual-priority federated pretraining

arXiv:2609.27611v1 Announce Type: new Abstract: Foundation models hold promise for generalizable analysis of structural brain magnetic resonance imaging (MRI) across development, aging and disease. Ho

applicationsarxiv-cs-cv

A Hierarchy-Aware Video-Language Model Evaluation and Hyperbolic Baseline for Surgery

arXiv:2609.27139v1 Announce Type: cross Abstract: Surgical procedures follow a phase-to-step hierarchy, yet the video-language models used to recognize them are evaluated with flat per-level metrics t

A hierarchy of faithfulness criteria for knowledge base completion

arXiv:2609.27863v1 Announce Type: new Abstract: Knowledge graph completion is evaluated by ranking observed triples above randomly corrupted ones, which treats every unobserved fact as false. When the

A Hybrid Iterative Deep Ritz Method for Elliptic Interface Problems

arXiv:2609.27325v1 Announce Type: cross Abstract: In this work, we propose a hybrid iterative deep Ritz method (H-IDRM) for a class of interface problems for second-order elliptic operators. It is bas

A Leakage-Aware Multimodal Evaluation Framework for Early Intraoperative Acute Kidney Injury Prediction

arXiv:2609.26848v1 Announce Type: cross Abstract: Postoperative acute kidney injury (AKI) after major non-cardiac surgery carries substantial morbidity, yet early intraoperative risk stratification re

A Modular Dual-Arm Robotic Cell for Disassembly and Repair of Industrial Control Electronics

arXiv:2609.27466v1 Announce Type: new Abstract: Industrial control electronics such as programmable logic controllers, servo drives and operator panels are routinely repaired in plant maintenance, but

researcharxiv-cs-ro

Tutorials

A Native-Reference Coordinate Geometry for L2 Pronunciation Deviation Using Self-Supervised Speech Models

arXiv:2609.28060v1 Announce Type: new Abstract: Self-supervised speech models encode rich phonetic information, but it remains unclear how to transform this information into interpretable metrics for

tutorialsarxiv-cs-cl

A new, no-compromises database architecture for the agentic era

model-releasesgoogle-cloud-ai

Entire database engineering careers have been spent on a single question: How do you scale an OLTP workload without compromising the system of record that owns the data? Exadata answered the question