Search: “open-weights”

arXiv:2608.20347v1 Announce Type: cross Abstract: Language models (LMs) often pass behavioral bias evaluations, but it remains unclear whether they no longer represent the underlying associations that

safetyarxiv-cs-ai

24 Aug 2026

Content type

All Blog X Post Paper YouTube Reddit GitHub

Clear filters

Safety

Auditing Cross-Lingual Fairness in Language Model Watermarking

DGX agent

arXiv:2608.20047v1 Announce Type: new Abstract: Watermarking schemes for large language model output are evaluated almost exclusively on English text using each scheme's detection threshold and a narr

safetyarxiv-cs-cl

21 Aug 2026

Agents

Calibrated Triage, Not Autonomy: Confidence Estimation for Medical Vision-Language Models

arXiv:2606.15910v2 Announce Type: replace Abstract: A vision-language model can answer a question about a chest radiograph or a pathology slide fluently and confidently while barely using the image, r

agentsarxiv-cs-cl

Compliance, Capability, and Conflict: Benchmarking Multimodal LLMs under System Messages

arXiv:2608.19207v1 Announce Type: new Abstract: Production deployments of Multimodal Large Language Models (MLLMs) increasingly rely on system messages to govern model behavior. Yet existing benchmark

FleetSieve: Decision-Critical Profiling for SLO-Aware LLM Fleet Configuration

model-releasesarxiv-cs-lg

arXiv:2608.19659v1 Announce Type: new Abstract: Choosing tensor-parallel (TP) degrees and replica counts for an LLM serving fleet is difficult because performance is not monotonic in TP and the feasib

Frequency-Aware Continual Learning for Smart Contract Vulnerability Detection with Large Language Models

arXiv:2608.19680v1 Announce Type: new Abstract: Smart contract vulnerability detection with Large Language Models (LLMs) faces three causally linked challenges. First, new vulnerability categories dem

OenoBench: A Wine-Domain Benchmark for Knowledge-Grounded Evaluation of Large Language Models

arXiv:2608.20106v1 Announce Type: new Abstract: We introduce OenoBench, a wine-domain knowledge benchmark of 3,266 multiple-choice questions across six pillars (regions, grape varieties, viticulture,

One Success Isn't Reliability: Thinkingbox, a Sandbox and Benchmark for Agents in Stateful Business Workflows

arXiv:2608.19741v1 Announce Type: new Abstract: Recent agent benchmarks increasingly ground evaluation in executable environments, from code repair to web navigation, app APIs, and function calling. Y

When Text and Numbers Disagree: Evidence Arbitration in Large Language Models

arXiv:2608.20116v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used in settings where textual summaries, numerical observations, and external tool outputs may provide co

Compress and Forget: bitsandbytes Quantization Amplifies Proactive Interference in LLMs

arXiv:2608.18578v1 Announce Type: new Abstract: Proactive interference (PI) is a documented failure mode in large language models in which retrieval of a repeatedly overwritten value degrades as prior

Evaluating Structured Information Extraction with Open Models in a High Risk Public Sector Application

arXiv:2608.18289v1 Announce Type: new Abstract: The extraction of structured information from unstructured documents represents a critical component of digital transformations in all sectors. While pr

Latent Space Refusal Anchoring for Low-Resource African Languages: Mechanistic Safety Recovery Without Retraining

arXiv:2608.18089v1 Announce Type: cross Abstract: Instruction-tuned models often refuse harmful requests in English but comply with the same requests in Yoruba, Igbo, Igala, and Hausa. This suggests t

Million-scale multimodal pollen microscopy with expert-guided foundation models

model-releasesarxiv-cs-cv

arXiv:2606.17809v2 Announce Type: replace Abstract: Automated pollen identification from microscopy remains a bottleneck in aerobiology, palaeoecology and biodiversity monitoring, because scalable sys

Safety Alignment Illusion: The Cross-Lingual Safety Gap in LLMs

arXiv:2608.18131v1 Announce Type: new Abstract: Current safety alignment training for Large Language Models (LLMs) are heavily English-centric. When such safety filters fail for non-English languages,

Local Ai

AISA: AI Safety Assistant Framework for Continuous Improvement of Highway Construction

arXiv:2608.17184v1 Announce Type: new Abstract: Job Safety Analysis (JSA) and pre-task planning can benefit from prior incident records, yet historical accident data is often stored as unstructured na

local-aiarxiv-cs-cl

BEAR-Bench: A Bilingual Enterprise and Academic Reasoning Benchmark for Multimodal Models

arXiv:2608.17895v1 Announce Type: cross Abstract: While Multimodal Large Language Models (MLLMs) have made significant strides in visual comprehension, their ability to reason about text-dense, profes

Cytoarchitecture in Words: Weakly Supervised Vision-Language Modeling for Human Brain Microscopy

arXiv:2602.23088v2 Announce Type: replace Abstract: Vision foundation models increasingly support interactive scientific workflows, but natural-language interaction requires coupling visual representa

researcharxiv-cs-cv

Agents

PROBE: Manipulation-Grounded Visual Question Answering with VLM Agents

arXiv:2608.17129v1 Announce Type: new Abstract: Vision-language Models (VLMs) excel at 2D grounding, spatial reasoning and agentic tool-based planning in static scenes. However, consider asking a home

agentsarxiv-cs-cv

Local Ai

SGHA: Evidence-Grounded Research Problem Discovery with Local Language Models

arXiv:2608.17501v1 Announce Type: new Abstract: Recent efforts toward fully automated AI scientists have demonstrated that language-model agents can generate hypotheses, execute experiments, and draft

local-aiarxiv-cs-ai

Which Source Wins? Task-Dependent Reliance in Vision-Language Models

arXiv:2608.17205v1 Announce Type: new Abstract: Vision-language models (VLMs) combine images and text, but when the two conflict and one becomes harder to read, it is unclear how a model shifts its re

Aborted but Not Forgotten: KV-Cache Retention Breaks Rollback Consistency in Language Agents

arXiv:2608.15939v1 Announce Type: new Abstract: Stateful language agents assume a rejected branch can be taken back by clearing it from the application transcript. We show this breaks when the serving

Architecture-Dependent Causal Transfer of Activation States Across Large Language Models

arXiv:2608.16347v1 Announce Type: new Abstract: Direct communication between AI systems relies on natural language as an intermediate layer, incurring encoding/decoding overhead, token cost, and laten

CBX-Bench: A Human-Aligned MLLM Council for Benchmarking Concept Bottleneck Model Explanations

model-releasesarxiv-cs-cv

arXiv:2608.15404v1 Announce Type: new Abstract: Concept Bottleneck Models (CBMs) are designed to make visual classification interpretable by expressing predictions through human-understandable concept

CEDAR-GRPO: Process-Aware Reinforcement Learning for General Abductive Reasoning in LLMs

arXiv:2608.14791v1 Announce Type: new Abstract: Abductive reasoning, often characterized as inference to the best explanation, is central to explanation under uncertainty, from everyday sense-making a

Decorrelation Is Not Complementarity: Skill, Not Lineage, Governs Trusted-Monitor Ensembles

arXiv:2608.16190v1 Announce Type: cross Abstract: Trusted monitoring has a cheap, trusted model score a stronger untrusted model's actions, and a diverse ensemble of them beats a single stronger monit

researcharxiv-cs-lg

Evaluating Multimodal LLMs across Text and Audio Modalities for Accessible Disaster Assistance

arXiv:2608.14651v1 Announce Type: new Abstract: Effective disaster risk communication is a foundational humanitarian challenge, yet current emergency infrastructure fails to meet the needs of individu

researcharxiv-cs-ai

Frontier AI Forecasting Has a Measurement Problem: An Audit of Progress Evidence

arXiv:2608.14903v1 Announce Type: new Abstract: Quantitative forecasts of frontier artificial intelligence often connect dated targets to trends in benchmark scores, training compute, release time, or

Gathered, Not Admitted: How Attention Brings a Latent Variable into Verbalizable Form

arXiv:2608.15022v1 Announce Type: new Abstract: Language models hold latent quantities in a form they can report on, and more of a quantity is present in that form when the task requires reusing it fl

Governance at the Boundary: How Agent Decomposition Degrades Policy Compliance

arXiv:2608.16055v1 Announce Type: new Abstract: Existing agent benchmarks ask whether the agent finished the task. We ask whether it finished it within policy. We introduce Fiducia-bench, a benchmark

Handoff-H1: An Orchestrated Vision-Agent System for Material Quantity Takeoff from Construction Blueprints

arXiv:2608.15032v1 Announce Type: cross Abstract: Converting a set of architectural blueprints into a complete material quantity takeoff requires visual perception across drawing sheets, dimensional a

L3Cube-IndicQuest v2: A Large-Scale Multilingual Benchmark for Evaluating Factual Knowledge of Large Language Models Across Indic Languages

arXiv:2608.15535v1 Announce Type: new Abstract: We present L3Cube-IndicQuest v2, a large-scale gold-standard multilingual question-answering benchmark for evaluating the India-specific factual knowled

LLMs for Zero-Shot Threat Detection via Structured Risk Indicators

model-releasesarxiv-cs-lg

arXiv:2608.16508v1 Announce Type: cross Abstract: We propose a two-stage large language model (LLM) framework for zero-shot detection of insider threats and advanced persistent threats (APTs) from het

Local Ai

Local AI pre-screening for human triple-blind peer review in health sciences

arXiv:2608.14625v1 Announce Type: cross Abstract: Academic peer review is under mounting strain: NeurIPS 2025 received 21,575 submissions, ICLR 2025 received 11,603, and ICML 2025 received 12,107. Thi

local-aiarxiv-cs-ai

One Score, Two Decisions: Selective Prediction on the Rare-Disease Tail

arXiv:2608.14683v1 Announce Type: new Abstract: Given a patient's clinical findings, a diagnostic system ranks possible diseases and must decide when to endorse its first prediction or defer it for re

researcharxiv-cs-lg

Applications

Plausible but Not Valid: A Psychometric Audit of LLMs as Synthetic Survey Respondents

arXiv:2608.14606v1 Announce Type: cross Abstract: Large language models (LLMs) are increasingly used as synthetic survey respondents, but existing evaluations ask whether answers look plausible at the

applicationsarxiv-cs-ai

Policy Iteration with Human Feedback: Bringing Post-Training RL to In-context Learning

arXiv:2608.16831v1 Announce Type: new Abstract: Generative pretraining established reusable task representations; later work on language-based task conditioning and in-context learning showed that a f

Safety

Revisiting Classifier-Free Guidance Methods in Latent Diffusion Models

arXiv:2608.16786v1 Announce Type: new Abstract: Inference-time quality-enhancement methods are an effective and widely adopted means of improving diffusion models without expensive retraining. We stud

safetyarxiv-cs-cv

Routing Divergence Is Not Evidence of Behavioral Influence in Same-Weight MoE Self-Distillation

arXiv:2608.15787v1 Announce Type: cross Abstract: Two Mixture-of-Experts (MoE) forward passes can share every weight yet route the same token through different experts. This creates a possible blind s

researcharxiv-cs-ai

The Recall Trap: A Recall-Maximizing Retriever Configuration Reduces Issue Resolution in Fixed-Budget Code Context

arXiv:2608.14838v1 Announce Type: cross Abstract: Retrieval components for code assistants are tuned against retrieval metrics: a configuration that raises recall@k is adopted, and downstream task suc

Agents

Workspace Topology as an Attack Vector in Agentic Coding Assistants

arXiv:2608.14876v1 Announce Type: cross Abstract: Agentic coding assistants are finding widespread use, not just in new code development but in quickly ingesting and leveraging third-party code. This

agentsarxiv-cs-ai

A Four-Axis Trustworthiness Benchmark for LLM-as-Judge in Principle-Based Regulation

arXiv:2608.14329v1 Announce Type: cross Abstract: Principle-based regulation, with evaluative standards such as 'fair, clear, and not misleading' or 'deliver good outcomes', cannot be reduced to binar

AnchorBench: A Multi-Pathway Benchmark for the Anchoring Effect in LLMs

arXiv:2608.14320v1 Announce Type: new Abstract: The anchoring effect is a cognitive bias in which an initial reference value shifts a later judgment toward itself. This effect is well established in h

ChartProbe: A Diagnostic Study on Visual Reasoning through Perception, Grounding, and Simple Reasoning

model-releasesarxiv-cs-cv

arXiv:2608.13766v1 Announce Type: new Abstract: Vision-language models (VLMs) remain unreliable on chart questions that require reasoning over visual quantities, and this weakness is usually attribute

Tutorials

CytoBERT: A Foundation Model for Cytometry Data

arXiv:2608.14414v1 Announce Type: new Abstract: Cytometry measures the complex characteristics of single cells (e.g., counts and protein expression of immune cells) and is widely used across immunolog

tutorialsarxiv-cs-lg

Detecting Contaminated Code-Generation Prompt Batches via Influence Functions

model-releasesarxiv-cs-lg

arXiv:2608.14303v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used for code generation, yet they remain vulnerable to prompts that elicit insecure implementations. Exis

Implementing Computational Law in Wolfram Language for the Governance of Artificial Intelligence

arXiv:2608.13958v1 Announce Type: new Abstract: How do we govern AI systems whose reasoning we cannot fully inspect? Governance does not require understanding a system's reasoning. It requires stating

Jais 2: A Family of Arabic-Centric Open Large Language Models

arXiv:2608.13580v1 Announce Type: cross Abstract: Jais 2 is a family of Arabic-centric large language models developed jointly by MBZUAI, Cerebras, and Inception, designed to advance Arabic-centric la

Repair, Not Improvement: Decomposing Constrained Decoding in Tool-Call Abstention

arXiv:2608.13959v1 Announce Type: new Abstract: Function calling is what the recent accounting of constrained generation explicitly sets aside: it finds the decoder's contribution small for format con

researcharxiv-cs-cl