AI news on Friday, June 12, 2026

SIGNIFICANCE

★ Top story · Models5d ago

Scammers Used Gemini AI to Help Build Spam Messages, Google Says

Google has filed a lawsuit against a group it identifies as Outsider Enterprise for allegedly using its Gemini AI to draft and distribute over two million fraudulent text messages. The operation reportedly targeted hundreds of thousands of users with deceptive links intended to facilitate financial scams. This legal action highlights the increasing concern among major tech firms regarding the exploitation of generative AI for large-scale cybercrime. By partnering with the FBI, Google aims to disrupt a sophisticated network that demonstrates how bad actors are weaponizing automated tools.

TThe Decoder BBloomberg Technology TTechCrunch AI Read original ↗

Friday — June 12, 2026

Scammers Used Gemini AI to Help Build Spam Messages, Google Says

Moonshot AI Launches Kimi Work, a Local Desktop Agent Reportedly Running on Kimi K2.6 With a 300-Sub-Agent Agent Swarm - MarkTechPost

Big Tech’s desperate last push at AI regulation

Google Research's Gemini-SQL2 tops text-to-SQL benchmarks by a wide margin

Anthropic, OpenAI, Google Executives Plan to Attend G7 Summit

Apple’s New Siri Is Just Good Enough to Ease Its AI Crisis

Results from the first Anthropic Public Record - Anthropic

Sarvam AI's Dr. Pratyush Kumar Headlines India-Nepal Seminar on AI Collaboration Convened by NICCI, Embassy of India, and AIDIA - nicci.org

How Preply combines AI and human tutors to personalize learning

The AI industry's platform trap is starting to look a lot like Microsoft's

NVIDIA Blackwell Leads on First Agentic AI Infrastructure Benchmark

Apple’s Camera Chief Thinks AI Can Give You Superpowers

The Next US-China AI Battle Is Over Compute &mdash; and China Is Spending Big

TCS and Anthropic partner to bring Claude to regulated industries - Anthropic

New OpenAI Academy courses for the next era of work

Latest AI Doomers Imagine How Europe ‘Slides Into Irrelevance’

Anthropic's Claude Fable 5 costs twice as much for 5.7 percent more performance

Multi-Turn Reasoning When Context Arrives in Pieces: Scalable Sharding and Memory-Augmented RL

SENTINEL: Failure-Driven Reinforcement Learning for Training Tool-Using Language Model Agents

When Does Delegation Beat Majority? A Delegation-Based Aggregator for Multi-Sample LLM Inference

Iterative Visual Thinking: Teaching Vision-Language Models Spatial Self-Correction through Visual Feedback

LLM-based Embeddings: Attention Values Encode Sentence Semantics Better Than Hidden States

PaLMR: Towards Faithful Visual Reasoning via Multimodal Process Alignment

Ex-Omni: Enabling 3D Facial Animation Generation for Omni-modal Large Language Models

DiskChunGS: Large-Scale 3D Gaussian SLAM Through Chunk-Based Memory Management

Rethinking RAG in Long Videos: What to Retrieve and How to Use It?

sebis at CRF Filling 2026: A Two-Stage Local LLM Pipeline for Medical CRF Filling

Brick: Spatial Capability Routing for the Mixture-of-Models (MoM) Paradigm

Possibilistic Predictive Uncertainty for Deep Learning

VISTA: An End-to-End Benchmark for Visual Spec-to-Web-App Coding Agents

Structuring The Future: Diffusion LLM Speculative Decoding via Calibrated Draft Graphs

NavWAM: A Navigation World Action Model for Goal-Conditioned Visual Navigation

Where, What, Why, and Importance: Structured Defect Grounding for Text-to-Image Feedback

Hierarchical Successor Representation for Robust Transfer

Muse Spark Safety & Preparedness Report

InterleaveThinker: Reinforcing Agentic Interleaved Generation

Different Layers, Different Manifolds: Module-Wise Weight-Space Geometry in Transformer Optimization

Select to Think: Unlocking SLM Potential with Local Sufficiency

Towards Data-free and Training-free Compression for Speech Foundation Models Using Parameter Clustering

A Unifying Lens on Reward Uncertainty in RLHF

X-MADAM-RAG: Diagnosing and Handling Chinese-English Evidence Conflict in Retrieval-Augmented Generation

HiGR: Industrial-Scale Hierarchical Generative Slate Recommendation Framework in Tencent

Metriplectic Conditional Flow Matching for Dissipative Dynamics

Zuckerberg says Meta made 'mistakes' in AI workforce shift

ReSET: Accurate Latency-Critical NVFP4 Reasoning via Step-Aware Temperature Scaling

Multi-Bitwidth Quantization for LLMs Using Additive Codebooks

DiffCoord: Differentiable Coordination for Distributed Multi-Agent Trajectory Optimization

DrivingAgent: Design and Scheduling Agents for Autonomous Driving Systems

When Will the UK Have Its First £100 Billion Tech Firm?

Mana: Dexterous Manipulation of Articulated Tools

RogueAI: A Reverse Turing Test for Detecting Licensed AI Deception in Dialogue

Goal2Pixel: Grounding Goals to Pixels for Vision-Language Navigation

Decoding the Multimodal Maze: A Systematic Review on the Adoption of Explainability in Multimodal Attention-based Models

MoVerse: Real-Time Video World Modeling with Panoramic Gaussian Scaffold

ReFree: Towards Realistic Co-Speech Video Generation via Reward-Free RL and Multilevel Speech Guidance

From Verdict to Process: Agentic Reinforcement Learning for Multi-Stage Fact Verification

Versioned Late Materialization for Ultra-Long Sequence Training in Recommendation Systems at Scale

AgentBeats: Agentifying Agent Assessment for Openness, Standardization, and Reproducibility

Selecting Samples on Graphs: A Unified Dataset Pruning Framework for Lossless Training Acceleration

Arbor: Tree Search as a Cognition Layer for Autonomous Agents

Epistemic Constitutionalism Or: how to avoid coherence bias

Google Launches Colab CLI for Developers, Automation, and AI Agents

Two Wrongs, No Right: Auditing Social-Desirability Bias in LLM Annotators for Computational Social Science

Masked and Predictive Self-Supervised Foundation Models for 3D Brain MRI

SupraBench: A Benchmark for Supramolecular Chemistry

MARD: Mirror-Augmented Reasoning Distillation for Mechanism-Level Drug-Drug Interaction Prediction

Uncertainty-Aware Hybrid Retrieval for Long-Document RAG

Fantastic Scientific Agents and How to Build Them: AgentBuild for Rietveld Refinement

Competition and Diversity in Generative AI

Optimal Spatio-Temporal Decoupling for Bayesian Conformal Prediction

VISA: VLM-Guided Instance Semantic Auditing for 3D Occupancy World Models

HD-Prot: A Protein Language Model for Joint Sequence-Structure Modeling with Continuous Structure Tokens

Operads for compositional reasoning in LLMs

Position: Generative Engine Optimization Creates Underexamined Risks, Governance Must Target Concentration, Disclosure, and Academic Blind Spots

GENIE: A Fine-Grained Measure for Novelty

From Isolation to Entanglement: When Do Interpretability Methods Identify and Disentangle Known Concepts?

Polar: A Benchmark for Evaluating Political Bias in LLMs

Reliability of Probabilistic Emulation of Physical Systems

Reward Modeling for Multi-Agent Orchestration

The Next US-China AI Battle Is Over Compute — and China Is Spending Big