Every paper checked — nothing hidden

All papers

Every paper checked for brief #4 of 12–19 Oct 2026 — nothing is hidden. Brief papers first.

All17,313 In brief24 More (on-topic)3,333 Not relevant13,754 Not assessed87 Older352 Could not confirm93 Blogs with no paper266 All blogs284

Paper	Topic	Relevance
OnTrack: Real-Time Monitoring and Intervention in LLM Agent Trajectories via Streaming Structure-Aware Optimal Transport Proposes a streaming optimal transport method to monitor LLM agent trajectories in real time against reference runs. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not focus on model compression, quantization, or efficient model serving.
OnTrack: Real-Time Monitoring and Intervention in LLM Agent Trajectories via Streaming Structure-Aware Optimal Transport Proposes a streaming optimal transport method to monitor LLM agent trajectories in real time against reference runs. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Works with agent execution logs and tool calls, not multimodal vision-language inputs.
OnTrack: Real-Time Monitoring and Intervention in LLM Agent Trajectories via Streaming Structure-Aware Optimal Transport Proposes a streaming optimal transport method to monitor LLM agent trajectories in real time against reference runs. Preprint	Retrieval-augmented generation (RAG) via arxiv · 12 Oct, 16:28	Not relevant Does not address document retrieval, chunking, or knowledge grounding mechanisms.
OnTrack: Real-Time Monitoring and Intervention in LLM Agent Trajectories via Streaming Structure-Aware Optimal Transport Proposes a streaming optimal transport method to monitor LLM agent trajectories in real time against reference runs. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Proposes an online monitoring mechanism rather than an evaluation benchmark.
ARC: A Reasoning Recipe for Robot Foundation Models Presents a method to improve zero-shot robot foundation model performance using causal reasoning traces generated from demonstrations. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Does not focus on safety guardrails, jailbreak prevention, or deployment alignment.
ARC: A Reasoning Recipe for Robot Foundation Models Presents a method to improve zero-shot robot foundation model performance using causal reasoning traces generated from demonstrations. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Addresses general robot manipulation policy learning rather than industrial visual inspection.
ARC: A Reasoning Recipe for Robot Foundation Models Presents a method to improve zero-shot robot foundation model performance using causal reasoning traces generated from demonstrations. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Presents a training and prompting recipe for robot models, not an evaluation benchmark.
ARC: A Reasoning Recipe for Robot Foundation Models Presents a method to improve zero-shot robot foundation model performance using causal reasoning traces generated from demonstrations. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Focuses on robot trajectory control via reasoning traces rather than tool-using LLM software agents.
GeoReform: Reflective Formalization Evolution for Multimodal Geometry Problem Solving Proposes an evolutionary framework that optimizes geometric diagram formalizations to improve multimodal geometry problem solving. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Focuses on math diagram reasoning accuracy rather than AI safety or reliability guardrails.
GeoReform: Reflective Formalization Evolution for Multimodal Geometry Problem Solving Proposes an evolutionary framework that optimizes geometric diagram formalizations to improve multimodal geometry problem solving. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Proposes a formalization framework to improve reasoning, not a new benchmark or evaluation methodology.
GeoReform: Reflective Formalization Evolution for Multimodal Geometry Problem Solving Proposes an evolutionary framework that optimizes geometric diagram formalizations to improve multimodal geometry problem solving. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Deals with geometry formalization in MLLMs rather than multi-step tool-using agents.
HRIL: Learning Multimodal Synergy via Higher-Order Tensor Modeling Proposes a higher-order tensor modeling framework to preserve synergistic cross-modal interactions in self-supervised multimodal representations. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies multimodal tensor representations, not model quantization, distillation, or efficient inference.
HRIL: Learning Multimodal Synergy via Higher-Order Tensor Modeling Proposes a higher-order tensor modeling framework to preserve synergistic cross-modal interactions in self-supervised multimodal representations. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Does not address AI safety, hallucination, guardrails, or robustness.
HRIL: Learning Multimodal Synergy via Higher-Order Tensor Modeling Proposes a higher-order tensor modeling framework to preserve synergistic cross-modal interactions in self-supervised multimodal representations. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Focuses on multimodal representation learning, not LLM agents or tool planning.
HRIL: Learning Multimodal Synergy via Higher-Order Tensor Modeling Proposes a higher-order tensor modeling framework to preserve synergistic cross-modal interactions in self-supervised multimodal representations. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Proposes a representation learning method rather than an evaluation benchmark.
SpaceFlow: Locally Controllable 3D Generation Introduces a training-free 3D generation pipeline enabling local geometric and appearance control using 3D primitives. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Presents a 3D shape generation pipeline, not an LLM agent executing tasks.
SpaceFlow: Locally Controllable 3D Generation Introduces a training-free 3D generation pipeline enabling local geometric and appearance control using 3D primitives. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on generative 3D shape modeling rather than computer vision for industrial inspection.
SpaceFlow: Locally Controllable 3D Generation Introduces a training-free 3D generation pipeline enabling local geometric and appearance control using 3D primitives. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Does not study AI safety, guardrails, or alignment.
SpaceFlow: Locally Controllable 3D Generation Introduces a training-free 3D generation pipeline enabling local geometric and appearance control using 3D primitives. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Addresses 3D shape synthesis from primitives, not general multimodal reasoning or vision-language understanding.
SpaceFlow: Locally Controllable 3D Generation Introduces a training-free 3D generation pipeline enabling local geometric and appearance control using 3D primitives. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Introduces a 3D generation technique rather than an evaluation benchmark.
LeWAM: A JEPA World Action Model with Diffusion-Steering-Based MPC Presents a bidirectional JEPA world action model with diffusion-steering model predictive control for robot state tracking and planning. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Focuses on a self-supervised world action model for control rather than an LLM agent planning with tools.
LeWAM: A JEPA World Action Model with Diffusion-Steering-Based MPC Presents a bidirectional JEPA world action model with diffusion-steering model predictive control for robot state tracking and planning. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Presents a general robotic world model rather than industrial inspection or defect detection.
LeWAM: A JEPA World Action Model with Diffusion-Steering-Based MPC Presents a bidirectional JEPA world action model with diffusion-steering model predictive control for robot state tracking and planning. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Develops a predictive model architecture and control framework, not an evaluation benchmark.
LeWAM: A JEPA World Action Model with Diffusion-Steering-Based MPC Presents a bidirectional JEPA world action model with diffusion-steering model predictive control for robot state tracking and planning. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Addresses latent dynamics prediction in robotics rather than vision-language multimodal foundation models.
LeWAM: A JEPA World Action Model with Diffusion-Steering-Based MPC Presents a bidirectional JEPA world action model with diffusion-steering model predictive control for robot state tracking and planning. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Does not evaluate or design deployment guardrails, safety alignment, or jailbreak defenses.
Searching for "Harmful Refusal": A Psychometric Audit of an AI Safety Benchmark Applies psychometric auditing to HELM Safety benchmarks to test whether harmful refusal functions as a single measurable attribute. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Analyzes benchmark validity for model safety refusals, not tool-using or autonomous agent systems.
Predicting Alignment Generalization with Value Representations Evaluates whether internal model representations can predict how fine-tuning on one value generalizes across other held-out values. Preprint	LLMs for code via arxiv · 12 Oct, 16:28	Not relevant Focuses on general value alignment, not code generation, review, or software engineering.
Predicting Alignment Generalization with Value Representations Evaluates whether internal model representations can predict how fine-tuning on one value generalizes across other held-out values. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Studies post-training alignment generalization rather than tool use or multi-step agent architectures.
Predicting Alignment Generalization with Value Representations Evaluates whether internal model representations can predict how fine-tuning on one value generalizes across other held-out values. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Work is restricted to text LLM alignment, not multimodal or vision-language models.
Predicting Alignment Generalization with Value Representations Evaluates whether internal model representations can predict how fine-tuning on one value generalizes across other held-out values. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not investigate model compression, distillation, quantization, or efficient serving.
Predicting Alignment Generalization with Value Representations Evaluates whether internal model representations can predict how fine-tuning on one value generalizes across other held-out values. Preprint	Retrieval-augmented generation (RAG) via arxiv · 12 Oct, 16:28	Not relevant Does not involve external document retrieval, chunking, grounding, or RAG architectures.
MAMHOI: Factorizing Scene-Aware Human-Object Interaction through Affordances Factorizes scene-aware human-object interaction generation into motion-affordance prediction and affordance-conditioned motion synthesis. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Proposes a generative synthesis model rather than an LLM evaluation benchmark.
MAMHOI: Factorizing Scene-Aware Human-Object Interaction through Affordances Factorizes scene-aware human-object interaction generation into motion-affordance prediction and affordance-conditioned motion synthesis. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Addresses 3D graphics and motion synthesis rather than cross-modal understanding in multimodal language models.
MAMHOI: Factorizing Scene-Aware Human-Object Interaction through Affordances Factorizes scene-aware human-object interaction generation into motion-affordance prediction and affordance-conditioned motion synthesis. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Deals with 3D generative modeling of human-object interactions rather than multi-step tool-using language model agents.
MAMHOI: Factorizing Scene-Aware Human-Object Interaction through Affordances Factorizes scene-aware human-object interaction generation into motion-affordance prediction and affordance-conditioned motion synthesis. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on indoor human motion generation rather than industrial vision, inspection, or defect detection.
MAMHOI: Factorizing Scene-Aware Human-Object Interaction through Affordances Factorizes scene-aware human-object interaction generation into motion-affordance prediction and affordance-conditioned motion synthesis. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Does not address AI safety, jailbreaking, red-teaming, or model reliability.
RoboRSI: Stable, efficient, and reusable robot self-evolution in complex real-world environments Presents a top-down skill refinement system for robot self-improvement through structured execution feedback and reusable modular skills. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Focuses on robotic skill learning and software self-improvement rather than computer vision for industrial inspection.
RoboRSI: Stable, efficient, and reusable robot self-evolution in complex real-world environments Presents a top-down skill refinement system for robot self-improvement through structured execution feedback and reusable modular skills. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant The core contribution is robotic capability refinement, not deployment safety, guardrails, or alignment.
RoboRSI: Stable, efficient, and reusable robot self-evolution in complex real-world environments Presents a top-down skill refinement system for robot self-improvement through structured execution feedback and reusable modular skills. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Proposes a system framework for self-evolution, not an LLM evaluation benchmark or assessment methodology.
Ecology of AI Agents: Collaboration Creates a Population Threshold for Takeoff Formulates a population dynamics model to study takeoff thresholds and self-replication risks in collaborative AI agents. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Presents a theoretical dynamics model, not an empirical evaluation benchmark.
Caught in the Act: Probes Effectively Detect Sabotage and Catch Unverbalized Deception Develops multi-layer linear probes on model activations to detect agent deception and sabotage across diverse scenarios. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not investigate model compression, distillation, or serving efficiency.
Caught in the Act: Probes Effectively Detect Sabotage and Catch Unverbalized Deception Develops multi-layer linear probes on model activations to detect agent deception and sabotage across diverse scenarios. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Focuses on internal probe mechanisms to detect deception rather than agent architectures.
Caught in the Act: Probes Effectively Detect Sabotage and Catch Unverbalized Deception Develops multi-layer linear probes on model activations to detect agent deception and sabotage across diverse scenarios. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Introduces a probe-based detection method rather than a general LLM evaluation benchmark.
Bi-FORK: Generative Modeling of High-Dimensional Bifurcating Systems Proposes a generative flow matching method with repulsion-guided sampling to model one-to-many bifurcation trajectories in physical systems. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Focuses on flow matching for physical bifurcating systems, not LLM agents.
Bi-FORK: Generative Modeling of High-Dimensional Bifurcating Systems Proposes a generative flow matching method with repulsion-guided sampling to model one-to-many bifurcation trajectories in physical systems. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not address small language models or inference compression techniques.
Bi-FORK: Generative Modeling of High-Dimensional Bifurcating Systems Proposes a generative flow matching method with repulsion-guided sampling to model one-to-many bifurcation trajectories in physical systems. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Not related to AI safety, guardrails, or alignment.
Bi-FORK: Generative Modeling of High-Dimensional Bifurcating Systems Proposes a generative flow matching method with repulsion-guided sampling to model one-to-many bifurcation trajectories in physical systems. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Presents a scientific generative framework rather than an LLM evaluation benchmark.
BrickBench: Evaluating Agentic Brick Design Introduces a benchmark and environment for evaluating coding agents on text-conditioned LEGO design with physical and discrete assembly constraints. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Does not focus on safety risks, red-teaming, or guardrails.
BrickBench: Evaluating Agentic Brick Design Introduces a benchmark and environment for evaluating coding agents on text-conditioned LEGO design with physical and discrete assembly constraints. Preprint	Computer vision for industrial applications via arxiv · 12 Oct, 16:28	Not relevant Studies text-conditioned LEGO assembly rather than industrial vision inspection, defect detection, or robotics.
BrickBench: Evaluating Agentic Brick Design Introduces a benchmark and environment for evaluating coding agents on text-conditioned LEGO design with physical and discrete assembly constraints. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Focuses on 3D discrete assembly and constraint validation rather than multimodal vision-language models.
From Reactive Containment to Proactive Assurance: Lessons from OpenAI, Anthropic, and Google Agent Security Incidents Analyzes security incidents from frontier AI agent evaluations and proposes a framework for proactive boundary assurance. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Presents a case study and security framework, not a new evaluation benchmark.
On the estimation and validity of AI time horizons---a statistical look at the METR plot Recomputes METR task time horizons using splines and item-response theory to examine the validity of AI difficulty curves. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not focus on model compression, quantization, or efficient serving.
On the estimation and validity of AI time horizons---a statistical look at the METR plot Recomputes METR task time horizons using splines and item-response theory to examine the validity of AI difficulty curves. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Provides a statistical analysis of an evaluation metric, not agent systems or architectures.
On the estimation and validity of AI time horizons---a statistical look at the METR plot Recomputes METR task time horizons using splines and item-response theory to examine the validity of AI difficulty curves. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Studies benchmark metric validity rather than deployment safety, alignment, or guardrails.
AutoIndexer: Flexible-Order Decoding by Training Causal Models on Chains of Edits Introduces a generative model that allows flexible post-hoc text editing through learned insertion, substitution, and deletion operations. Preprint	Multimodal models via arxiv · 12 Oct, 16:28	Not relevant Deals with text token sequence generation, not multimodal models.
AutoIndexer: Flexible-Order Decoding by Training Causal Models on Chains of Edits Introduces a generative model that allows flexible post-hoc text editing through learned insertion, substitution, and deletion operations. Preprint	LLM evaluation and benchmarks via arxiv · 12 Oct, 16:28	Not relevant Presents a generative modeling technique rather than an evaluation framework.
AutoIndexer: Flexible-Order Decoding by Training Causal Models on Chains of Edits Introduces a generative model that allows flexible post-hoc text editing through learned insertion, substitution, and deletion operations. Preprint	Retrieval-augmented generation (RAG) via arxiv · 12 Oct, 16:28	Not relevant Focuses on generative decoding mechanisms, not retrieval-augmented generation.
AutoIndexer: Flexible-Order Decoding by Training Causal Models on Chains of Edits Introduces a generative model that allows flexible post-hoc text editing through learned insertion, substitution, and deletion operations. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on edit flexibility rather than model compression, quantization, or serving efficiency.
AutoIndexer: Flexible-Order Decoding by Training Causal Models on Chains of Edits Introduces a generative model that allows flexible post-hoc text editing through learned insertion, substitution, and deletion operations. Preprint	AI agents via arxiv · 12 Oct, 16:28	Not relevant Proposes a decoding mechanism for sequence editing, not multi-step autonomous agents.
AutoIndexer: Flexible-Order Decoding by Training Causal Models on Chains of Edits Introduces a generative model that allows flexible post-hoc text editing through learned insertion, substitution, and deletion operations. Preprint	AI safety and reliability in deployment via arxiv · 12 Oct, 16:28	Not relevant Not focused on safety, security, jailbreaks, or guardrails.

13,201–13,260 of 13,754← Previous Next →