AI news on Tuesday, June 23, 2026

SIGNIFICANCE

★ Top story · Models5d ago

Anthropic Wants Claude to Be Your New Slack Coworker

Anthropic is expanding its Claude AI integration within Slack to provide users with more advanced capabilities for managing workplace communications. By embedding the model more deeply into the platform, the company aims to streamline tasks like drafting messages and summarizing lengthy conversation threads. This update follows a broader industry trend of integrating generative AI directly into enterprise collaboration tools to improve professional productivity.

TThe Decoder BBloomberg Technology TTechCrunch AI Read original ↗

Tuesday — June 23, 2026

Anthropic Wants Claude to Be Your New Slack Coworker

Mistral OCR 4 : SOTA OCR for Document Intelligence

Google Poised to Lose Two More Senior AI Staffers to Anthropic

Introducing Claude Tag

How GPT-5 helped immunologist Derya Unutmaz solve a 3-year-old mystery

Washington’s Foreign Ban on Anthropic’s Top Models May Backfire

How Omio is building the future of conversational travel

ByteDance's Seedance 2.5 breaks the 30-second barrier for AI video generation

Abu Dhabi’s MGX Raises About $50 Billion to Accelerate AI Deals

Cursor announces its own AI model, a new Git platform, and a mobile app

How Meta Engineered Ultra-Narrow Batteries for AI Glasses

Helping build shared standards for advanced AI

B[FM]$^2$: Brain Foundation Model via Flow Matching with SplitUNet

SOHET: Sequence Of Heterogeneous Events Transformer with Self-Supervised Pre-Training

Darwin Mobile Agent: A Roadmap for Self-Evolution

Beyond Templates: Revisiting Zero-Shot Remote Sensing through Meta-Prompting

Phonemes to the Rescue: Multilingual Tokenization Based on International Phonetic Alphabet

Confidence Laundering in Agent Systems: Why Uncertainty Needs a Latent Carrier

What Shapes Emergent Misalignment? Insights from Training Dynamics, Model Priors, and Data

Specific Domain Ontology Construction Using Large Language Models

VeriBound: PAC-Bayesian Generalization Bounds for Process Reward Models Trained with Formal Verification Tools

NeoJaundice-AI: Smartphone-Based Neonatal Jaundice Detection Using Dual-Input Deep Learning and Synthetic Augmentation

Beyond Hooking Onto the World: Referential Profiles and the Numerical Structure of LLM Grounding

Negative Knowledge as Failure-aware Shared Memory for AutoResearch

Right Knowledge, Wrong Answer: Test-Time Steering for Temporal Fact Conflicts in Open-Weight Language Models

Go-with-the-Track: Video Compositing and Motion Control with Point Tracking

DataClaw0: Agentic Tailoring Multimodal Data from Raw Streams

Beyond Fixed Budgets: Characterizing the Inelasticity and Limitations of Tree-of-Thought Reasoning Strategies

LEViL: Label-Efficient Video Learning via Zero-Shot Distillation over VLM-Generated Pseudo-Label Spaces

Latent Personal Memory: Represent personal memory as dynamic soft prompts

Fine-grained Human Motion Understanding with Language Models

SEED: Simple ViT and Evolving Harness for Explainable Text Forgery Detection

Process-Reward Tactic Evolution for Long-Horizon Bioinformatics Workflows

ARGUSTRACK: A Multi-View Annotation System for Multi-Object Tracking

FiLM-Coordinated Dual-Branch Transformer for Global-Local Dependency Modeling in Language Modeling

Repeated post-training is not Self-improving: Diagnosing Scientific Amnesia in Continual DPO Pipelines

From Question Answering to Task Completion: A Survey on Agent System and Harness Design

Answer Engineering: Local Trajectory Editing for Protocol-Constrained Decision Making in Large Language Models

RIZZ: Routing Interactions to Near Zero-Interference Zones for Continual Adaptation of Black-Box Agents

Post-Training Recipe, More Than Model Family, Shapes Multi-Agent LLM Conversational Behavior

ELDiff: When Evidential Learning Meets Text-to-Image Diffusion

PROTON: Prototype-Based Test-Time Online OOD Detection for Medical VLMs

From Uncertainty to Stability and Fidelity: Guiding Sparse-View 3D Gaussian Splatting with Fisher Information

Beyond 'One Language, One Script': Quantifying Orthographic Bias in Multilingual VLMs with PuMVR

When Web Agents Finish but Still Fail: Reproducible Triggers and Trace Diagnostics for Parallel Web Exploration

AEF-Econ: Toward Plug-and-Play Socioeconomic Foundation Embeddings from AlphaEarth for Urban Remote Sensing

An LLM-Explainable DRL Framework for Passenger-Directed Autonomous Driving

AlphaMemo: Structured Search-Process Memory for Self-Evolving Alpha Mining Agents

Chem2Gen-Bench: Benchmarking Chemical-to-Genetic Translation in Perturbation Response Space

Is Our Benchmark Enough? An Analysis of Continual Learning for MLLMs

When Compression Helps and When It Hurts: Condition-Aware Analysis of Chain-of-Thought Distillation

Neurosymbolic Clinical Trial Matching via LLM-Driven Abduction and Logical Verification

Hypothesis-Disciplined Multi-Agent Automated Formalization of Asymptotic Statistical Theory

Metric-Dependent Annotation Saturation for Learning from Label Distributions

SARIF: Segment Anything for Robust Image Forensics

Peeking Inside LLMs: Leveraging Internal Artifacts of LLMs for Enhancing Reliability in Legal Classification

Learning the ARTS of Search for Automated Discovery

On the Identifiability of User Adaptation in Co-Adaptive Neural Interfaces

Lexical Consensus: Grounded Word Learning and Shared Meaning in Artificial Agents

The AI Evaluability Gap: The Missing Layer for Managing Risk and Sustaining Value

CIExplainer++: Generating Causal and Interpretable Explanations for Graph Neural Networks

Skill Coverage: A Test Adequacy Metric for Agent Skills

Who Checks the Citations? Benchmarking Legal Hallucination Detection

Learning Burst-Aware Early Warning Models for Capacity Stress under AI Workload Surges in Hyperscale Data Centers

CogniRoute: Learning to Route Social Evidence in Omni-Modal Models

DCD-PFN: A Decoupling-Aware Foundation Model for Causal Discovery

Factual Retrieval in LLMs Is a Redundant, Distributed and Non-Contiguous Process

Extraction and Analysis of Multimodal Concepts in Vision Language Models through Sparse Autoencoders

SCOPE: Sequential Conformal Probing for Reliable OOD Rejection in LLM Services

ARCO: Adaptive Rubric with Co-Evolution for Multi-Step LLM-Based Agents

Inverting the Bellman Equation: From $Q$-Values to World Models

Dementia-Agents: A Multi-Modal Multi-Agent System for Dementia Staging and Phenotyping

Massive Activations Are Architecturally Robust: A Controlled Scratch/Commitment Residual Stream Test

A-Evolve-Training: Autonomous Post-Training of a 30B Model

Self-Improvement Can Self-Regress: The Rise-and-Collapse Failure Mode of LLM Self-Training

ChainWorld: Composing Long-Horizon Desktop Workloads from Atomic OSWorld Tasks

Scalable Hierarchical Attention Transformers for Multi-Turn Jailbreak Detection in Long Conversations

Neural Architecture Distributions: A New Paradigm for Stochastic Segmentation

Sakana Fugu Technical Report