AI news on Tuesday, June 30, 2026

SIGNIFICANCE

★ Top story · Models1d ago

Claude Science, an AI workbench for scientists, is now available

Anthropic has launched Claude Science, a specialized workbench designed to centralize computational research tasks within a single environment. By integrating disparate databases, analytical pipelines, and software tools into one platform, the tool aims to automate repetitive aspects of the research process. This release addresses significant productivity bottlenecks for scientists who previously spent considerable time navigating fragmented workflows. The shift suggests a broader trend of AI developers creating domain-specific environments to accelerate technical discovery and minimize administrative overhead in professional research settings.

AAnthropic BBloomberg Technology MMIT Technology Review Open full story →Read original ↗

Tuesday — June 30, 2026

Claude Science, an AI workbench for scientists, is now available

Anthropic Says US Lifted Restrictions on Fable 5

Anthropic's new Claude Sonnet 5 closes the gap to the pricier Opus model series

Introducing Claude Sonnet 5

Meituan's LongCat-2.0 shows China can train massive AI models without Nvidia

Meta's non-invasive brain-to-text AI is closing the gap with surgical implants

Start building with Nano Banana 2 Lite and Gemini Omni Flash

Google launches Nano Banana 2 Lite for fast AI images and Gemini Omni Flash for video via API

Hidden code in Claude Code secretly flagged Chinese users

Unlocking Britain’s next era of productivity: Building a nation of AI trailblazers

BOE’s Breeden Warns AI Agents Risk Triggering Market Meltdowns

How NVIDIA’s Inference Software Stack Powers the Lowest Token Cost

Aspiring Nvidia Rival Gets Backing From Jane Street

How Anthropomorphic Language Impacts Public Perceptions of AI

DataComp-VLM: Improved Open Datasets for Vision-Language Models

Vision-driven Preference Synthesis for Mitigating Hallucinations in VLMs

GPC: Large-Scale Generative Pretraining for Transferable Motor Control

DiffRGD: An Inference-Time Diffusion Guidance Through Riemannian Gradient Descent

MedEvoEval: Evaluating Continual Evolution of Doctor Agents through Simulated Clinical Episodes

Recursive Self-Evolving Agents via Held-Out Selection

SEAD: Competence-Aware On-Policy Distillation via Entropy-Guided Supervision

Meshtryoshka: Differentiable Rendering of Real-World Scenes via Mesh Rasterization

A Trainable-by-Parts Operator Learning Framework: Bridging DeepONet and Karhunen-Loeve Expansions for Large-Scale Applications

BayesEvolve: Explicit Belief States for Autonomous Scientific Discovery

Data Provenance for Image Auto-Regressive Generation

Multi-Block Diffusion Language Models

Anthropic launches AI drug discovery program, joining tech giants in betting on healthcare

Towards Evaluating Data Priors for Tabular Foundation Models

Self-Evolving Agentic Image Restoration via Deliberate Planning and Intuitive Execution

JuZhou 1.0 Technical Report: The First Edge-Native Text-to-Image Foundation Model Trained Entirely on China-Developed AI Accelerators

MammoFlow: Multiview Mammogram Synthesis with Anatomically Consistent Flow Matching

Agent Safety Is Action Alignment

ViPSim: Collaborating Visual and Parameter Spaces for Consistent Long-Horizon Embodied World Models

BV-Blend: Uncertainty-Weighted Historical Baselines for Stable Critic-Free RL with Verifiable Rewards

LLM agents security duality: a comprehensive survey of self-security and empowered cybersecurity

PASTA: A Paraphrasing And Self-Training Approach for Knowledge Updating in LLMs

Preventing Error Propagation in Multi-Agent AI through Runtime Monitoring

Personalizing MLLMs via Reinforced Multimodal Reference Game

Why Specialization Is Inevitable

Labeling Training Data for Entity Matching Using Large Language Models

Majority Vote Silences Minority Values: Annotator Disagreement at the Hate/Offensive Boundary in HateXplain

A Path-Space Formulation of Prediction in World Models: From a Single Action to Prediction, Planning, and Irreversibility

Emily Bender Sets the Record Straight on “Stochastic Parrots”

AEGIR: Modeling Area Emitters for Indoor Inverse Rendering using Gaussian Splatting

What LLMs explain is not what they believe: Evaluating explanation sufficiency under models' own input beliefs

Correct codes for the wrong reasons? validating LLMs as measurement instruments for theoretical constructs

Into the Omniverse: Three Workflows for Improving Vision AI Agent Accuracy With Synthetic Data and Fine-Tuning

Microsoft builds a bouncer to keep bots out of Teams meetings

SemFlowRAG: Directed Semantic Flow from Abstraction to Evidence for Complex Reasoning

OpenAI reportedly cut response costs for guest ChatGPT users by more than half

RSGPNet: Geometric Prompting for Remote Sensing Open-Vocabulary Semantic Segmentation

RADIANT-PET: Reasoning-Augmented PET/CT Lesion Segmentation with Large Language Models and Reinforcement Learning

Managing the Human Fallback: Skill Investment Under Improving AI and Worker Mobility

Invariant Reasoning Directions in Latent Trajectories of Language Models

Few-Step Boltzmann Generators via Scalable Likelihood Flow Maps

DLGStream: Dynamic Language-embedded Guassian Splatting for Open-vocabulary Enabled Free-viewpoint Video Streaming

New attack provides one more reason why AI browsers are a bad idea

Position: RL Researchers Need to Distinguish Between Solving Simulators and Using Simulators as a Proxy

Mechanistic Personality Analysis of LLMs Steering Personality via Latent Feature Interventions

US campaigns now run on AI at nearly every step, and Europe is drawing a harder line

When Is a Draft Accepted? A Theory of Acceptance in Speculative Decoding

GPTNT: Benchmarking Real-Time Collaboration Between Multimodal Agents on Keep Talking And Nobody Explodes

Symbolic Mechanistic Data Attribution: Tracing Training Influence to Learned Behavioral Policies

MIThinker: A Plug-and-Play Policy-Optimized Thinker For Motivational Interviewing Counseling

When More Sampling Hurts: The Modal Ceiling and Correlation Ceiling of Test-Time Scaling

When Summaries Distort Decisions: Information Fidelity in LLM-Compressed Financial Analysis

GLACIER: Rethinking Mass Spectrum Prediction as an Object Detection Problem

Agriculture is ready for AI, but its data isn’t

Depth Exploration for LLM Decoding

BaRA: Bayesian Adaptive Rank Allocation for Parameter-Efficient Fine-Tuning

Google’s NotebookLM can sum up your research in a TikTok-style clip

AI Notetakers in Meetings Raise Mounting Privacy Concerns

How Jaiveer Singh Is Helping Robots — and Developers — Move Faster

Beyond Backscatter: AlphaEarth Land-Cover Priors for Rapid SAR Flood Segmentation Across Foundation Backbones

LEDGER: Scaling Agentic Document Editing with Dependency-aware Graph Retrieval

AB-RAG: Adaptive Budgeted Retrieval-Augmented Generation for Reliable Question Answering

Semantic-Aware, Physics-Informed, Geometry-Grounded Weather Video Synthesis

Can LLMs Hire Fairly? Racial Bias in Resume Screening

Trump to lift limits on Anthropic’s Fable model