Daily feed

Top AI papers for today

Ranked by LLM judgment, embedding relevance, recency, and cached scoring signals.

Updated Apr 14, 2026

#10.754

Deliberative Alignment is Deep, but Uncertainty Remains: Inference time safety improvement in reasoning via attribution of unsafe behavior to base model

This research is significant as it tackles the persistent issue of unsafe behavior in language models, proposing a novel method to improve safety without compromising utility, which is crucial for the deployment of AI systems in sensitive applications.

alignmentbenchmarkevaluation

#20.747

GeoMeld: Toward Semantically Grounded Foundation Models for Remote Sensing

This paper addresses the challenge of semantically grounded foundation modeling in remote sensing by providing a large-scale dataset and a novel pretraining framework, potentially enhancing the accuracy and applicability of remote sensing models.

alignmentdatamultimodal

#30.745

Collaborative Multi-Agent Scripts Generation for Enhancing Imperfect-Information Reasoning in Murder Mystery Games

This paper addresses the challenge of improving VLMs' reasoning abilities in complex, adversarial environments, which is crucial for advancing AI's understanding and interaction in real-world scenarios with incomplete information.

agentbenchmarkevaluation

Full Top 10

#4AI Integrity: A New Paradigm for Verifiable AI GovernanceAI Integrity offers a novel approach to AI governance by ensuring the verifiability of reasoning processes.0.742 #5PRISM Risk Signal Framework: Hierarchy-Based Red Lines for AI Behavioral RiskPRISM proposes a hierarchy-based framework for AI risk detection, enhancing anticipatory safety measures.0.742 #6The Blind Spot of Agent Safety: How Benign User Instructions Expose Critical Vulnerabilities in Computer-Use AgentsThe paper uncovers significant vulnerabilities in CUAs under benign user instructions, emphasizing the need for improved safety measures.0.742 #7Instructing LLMs to Negotiate using Reinforcement Learning with Verifiable RewardsThe paper innovatively applies RLVR to enhance LLM negotiation capabilities.0.739 #8Thinking Fast, Thinking Wrong: Intuitiveness Modulates LLM Counterfactual Reasoning in Policy EvaluationThe paper highlights how intuitiveness influences LLMs' counterfactual reasoning in policy evaluation.0.739 #9Agentic Application in Power Grid Static Analysis: Automatic Code Generation and Error CorrectionA novel LLM agent automates power grid analysis with high accuracy and error correction.0.739 #10TorchUMM: A Unified Multimodal Model Codebase for Evaluation, Analysis, and Post-trainingTorchUMM provides a standardized framework for evaluating multimodal models.0.738