Latest AI research papers and analysis — helucino

37m
Advancing the price-performance frontier with GPT‑5.6
Simon Willison
7h
Echoverse: Deep, evolving environments for computer-use agents
Microsoft Research
9h
llm 0.32rc1
Simon Willison
13h
Ontologies Are So Back: Why AI Agents Are Reviving the Semantic Web
Latent Space
20h
Guarding Organizations Against Malware Risk: A Novel Graph-Based Malware Detection Method
arXiv cs.AI
20h
From Representations to Behaviors: Exploring the Person-Situation-Behavior Triad in LLMs
arXiv cs.AI
20h
MEDIC: Comprehensive Evaluation of Leading Indicators for LLM Safety and Utility in Clinical Applications
arXiv cs.AI
20h
Evidence-Ledger Adjudication for Claim-Evidence Traceability
arXiv cs.AI
20h
Automorphism-Induced Non-Canonicity in Top-k Explanations of Graph Neural Networks
arXiv cs.AI
20h
Dual Inversion for Text-to-Image Diffusion Models: From Both Prompt and Noise Perspectives
arXiv cs.AI
20h
Anatomy Contextualized Adaption of CT Foundation Models
arXiv cs.AI
20h
Cognitive Convergence: Deep Similarities Between Large Language Models and Human Cognition
arXiv cs.AI
20h
SkillRise: Agentic Reinforcement Learning for Cross-Task Skill Evolution
arXiv cs.AI
20h
Crossing-Free Probabilistic K-Line Forecasts Without Retraining
arXiv cs.AI
20h
The Capability Paradox: How Smarter Auditors Make Multi-Agent Systems Less Secure
arXiv cs.AI
20h
A Graph-Native Bitemporal Memory Store for Conversational AI Agents
arXiv cs.AI
20h
TANDEM: Temporal-Aware Neural Detection for Multimodal Hate Speech
arXiv cs.AI
20h
AI as Friction for Reflection Support in Ideation
arXiv cs.AI
20h
Partner Capability Estimation for Task-Agnostic Adaptation in Ad-Hoc Teamwork
arXiv cs.AI
20h
ResearchArena: Evaluating Sabotage and Monitoring in Automated AI R&D
arXiv cs.AI
20h
Recover, Decode, Reguard: Guard-Agnostic Defense Amplification againstEncoded VLM Jailbreaks
arXiv cs.AI
20h
Misalignment Has a Personality: A Big Five Account of Emergent Misalignment
arXiv cs.AI
20h
CoCaRS: Correlation Calibration-Based Redundancy Suppression for Heterogeneous Knowledge Distillation
arXiv cs.AI
20h
Do Latent Channels Actually Communicate? A Causal Audit of Latent Multi-Agent LLM
arXiv cs.AI
20h
Do Models Fake Alignment Without Clear Consequences?
arXiv cs.AI
20h
A Density-Matrix Framework for Electronic-Structure Analysis of Functional-Group and Salt Effects in Lithium-Metal Electrolytes
arXiv cs.AI
20h
Decision-oriented joint optimization of evidence fusion based on event-conditioned credibility
arXiv cs.AI
20h
Identifying Implicit Bias in LLM-based Chat AI Toward People with Intellectual Disabilities
arXiv cs.AI
20h
PLATO: Pointer Learner for Agent and Task Openness
arXiv cs.AI
20h
Can AI agents conduct open-ended AI research? Early evidence from two case studies
arXiv cs.AI
20h
RedKnot: Efficient Long-Context LLM Serving with Head-Aware KV Reuse and SegPagedAttention
arXiv cs.AI
20h
MemSecBench: Tracking Agent Memory Poisoning from Persistence to Consequence and Repair
arXiv cs.AI
20h
EvoPINN: Agentic Discovery of Executable Algorithms for Physics-Informed Neural Networks
arXiv cs.AI
20h
Living-Harness Is an Interactive-Agent Evolver
arXiv cs.AI
20h
CaM-Wolf: Causal-Aware Multimodal Agents for Social Deduction Games
arXiv cs.AI
20h
Collaborative Weighting with Pessimistic Critic for Mitigating Overestimation in Off-Policy Reinforcement Learning
arXiv cs.AI
20h
Reinforcement Learning on Cost-Constrained Quadrupedal Hardware
arXiv cs.AI
20h
On the Hybrid Nature of ABPMS Process Frames and its Implications on Automated Process Discovery
arXiv cs.AI
20h
Setoka: A Benchmark for Hierarchical User Understanding in Personalized Agents over Heterogeneous Data
arXiv cs.AI
20h
The Age of AI Agents Demands A New Scientific Paradigm To Sustain Trustworthy Science
arXiv cs.AI
20h
See2Think: Do Multimodal Models Really Use Intermediate Visual States?
arXiv cs.AI
20h
BioPro: Towards Difference-Aware Gender Fairness for Vision-Language Models
arXiv cs.AI
20h
A First Look at Coding Agents' Compliance with AI Contribution Rules in Open-Source Communities
arXiv cs.AI
20h
How does downsampling affect needle electromyography signals? A generalisable workflow for understanding downsampling effects on high-frequency time series
arXiv cs.AI
20h
A Picture Says Thousands of Words - Harnessing Dermal Exposure Data from Images through Hybrid Deep Learning for Enhanced Safety Assessment
arXiv cs.AI
20h
Budget-Aware LLM Discovery via Cost-Calibrated Frontier Utility
arXiv cs.AI
20h
Predict before you train: Scaling Laws for particle physics foundation models
arXiv cs.AI
20h
Do Methods Support the Claims? Intra-Paper Verification for Peer Review
arXiv cs.AI
20h
SkillSight: Calibrating Generic Content Bias for Skill Retrieval
arXiv cs.AI
20h
When LLMs Agree, Are They Right? Auditing Self-Consistency and Cross-Model Agreement as Confidence Signals
arXiv cs.AI
20h
ForgetBench: Benchmarking Forgetting Dynamics of Long-Term Parametric Memory in Language Models
arXiv cs.AI
20h
TraceCLIP: Recovering Local Semantics from Patch-to-CLS Contributions
arXiv cs.AI
20h
Borrowed Strength: Best-of-N Search over a Code EncodingBreaks Self-Check Jailbreak Defenses
arXiv cs.AI
20h
Ratchet: A Minimal Hygiene Recipe for Self-Evolving LLM Agents
arXiv cs.AI
20h
WhisperRec: Latent Reasoning for Efficient Foundation Recommendation Models
arXiv cs.AI
20h
High-Order Markov Blanket Discovery via a k-Order Relaxation of the Faithfulness Assumption
arXiv cs.AI
20h
One Run Is Not an Idea: The Implementation Lottery in Automated Research
arXiv cs.AI
20h
UrbanDS: A Graph-Guided LLM Multi-Agent System for Data-Intensive Urban Tasks
arXiv cs.AI
20h
Interactive Reward Agent: GUI Task Evaluation via Environment-State Verification
arXiv cs.AI
20h
HiFloat4 Format for End-To-End Reinforcement Learning Post-Training of Large Language Models
arXiv cs.AI