Agentic AI surges; NVIDIA expands factory partnerships

Anthropic / Claude ecosystem

Anthropic Launches Claude Fable 5 and Claude Mythos 5, Its First Public Mythos-Class Models

Anthropic has released Claude Fable 5, a Mythos-class model with built-in safeguards, to the public, alongside Claude Mythos 5 for select partners. The models introduce tiered access with advanced coding, research, and reasoning capabilities. Anthropic also announced Project Glasswing, deploying Mythos 5 to US government defenders.

Frontier model providers

Meta to Personalize Feeds Using Off-Platform Data with Global Privacy Toggle

Meta announced it will use off-platform data to improve feed and AI personalization across Instagram and Facebook, while adding a global privacy toggle for control. The move aims to enhance ad targeting and user experience.

Meta Removes Unreleased Face-Recognition From Smart Glasses After Report

Following a WIRED investigation, Meta deleted an unreleased face-recognition feature from its smart glasses companion app. The feature had been quietly integrated, raising privacy concerns.

Google Research Enhances Gemini Enterprise with Agentic RAG and Sufficient Context Agent

Google introduced Cross-Corpus Retrieval in the Gemini Enterprise Agent Platform, using a Sufficient Context Agent to improve multi-hop, multi-source search accuracy by up to 34%.

DeepSeek Launches Desktop GUI with Local AI Agents

DeepSeek released a desktop graphical user interface (GUI) that runs local AI agents for tasks like coding, writing, and automation, targeting enterprise messaging integration.

NotebookLM Upgraded with Gemini 3.5 Backend, Secure Cloud Computer, and Expanded Skills

Google upgraded NotebookLM with a Gemini 3.5 backend, a secure cloud computer, over 100 software skills, and new output formats, transforming the research assistant into a more powerful agent.

AI developer tooling & infrastructure

New Relic Introduces AI Coding Observability for Software Development

New Relic launched an open-source tool to provide observability into AI-assisted software development workflows, aiming to improve visibility and auditing of AI-generated code.

PointFive Launches AI Efficiency OS TokenShift to Optimize Cloud and Coding Agents

Startup PointFive introduced an AI Efficiency OS, TokenShift, that spans cloud and coding agents to reduce waste and improve AI/cloud spending efficiency.

LangChain Adds A2A Integration to Enable Multi-Agent Interoperability

LangChain introduced A2A-first integration, allowing any LangChain agent to expose as an A2A server and call other A2A-compliant agents across frameworks, reducing cloud lock-in.

Cloud & platform providers

Cloudflare Builds Frontier AI Defense Architecture as Customer Zero

Cloudflare detailed its defensive stack against frontier AI models, including an enhanced WAF scoring engine and validated-traffic API models, deployed on its own network first.

Cloudflare Introduces Real-Time Threat Intelligence into WAF Rules

Cloudflare added cf.intel threat indicators directly into its Web Application Firewall rule engine, enabling proactive mitigation based on real-time signals.

AI policy, regulation & governance

Senate Judiciary Committee Schedules Markup for AI Likeness Bill NO FAKES Act

The U.S. Senate Judiciary Committee set a markup for June 11 on S.4591, the NO FAKES Act, which addresses AI-generated likenesses, along with seven judicial nominations.

Industry & market moves

S&P Global Partners with Cohere to Power Trusted Agentic Workflows for Finance

S&P Global integrated its financial data into Cohere's North enterprise AI platform, enabling citation-backed, verifiable information in agentic workflows for financial institutions.

CATL Invests $1 Billion in DeepSeek AI Amid Battery Project

Chinese battery giant CATL invested $1 billion in DeepSeek as part of a larger AI funding round, while also supplying 2.4 GWh of battery storage to Australia. The investment underscores CATL's AI diversification.

NVIDIA and SK hynix Form Multiyear Partnership for AI Factory Memory

NVIDIA and SK hynix announced a multiyear technology partnership to co-develop next-generation memory solutions tailored for AI factories, aiming to accelerate global AI infrastructure.

NVIDIA and LG Group Forge Six-Company AI Factory Partnership

NVIDIA and LG Group announced a broad collaboration spanning six LG companies to build an AI factory, encompassing robotics, autonomous driving, data center cooling, and a sovereign AI model. The deal was facilitated by Jensen Huang's daughter.

AI product & feature launches

Picus Becomes Anthropic-Verified, Using Mythos-Class AI for Autonomous Pentesting

Picus Security announced it is now Anthropic-verified, integrating Mythos-class AI into its autonomous penetration testing platform. The verification provides frontier-grade cyber capabilities to defenders under Anthropic's program.

Alibaba's Amap Unveils ABot-Earth0.5: A 3D Native World Model for Cities

Alibaba subsidiary Amap released ABot-Earth0.5, a 3D native world model that can generate city-scale models from satellite images in 10 minutes on consumer GPUs, at a fraction of traditional costs.

MetaMask Launches AI Agent Wallet with Built-In Security for Crypto Trades

MetaMask introduced an AI agent wallet featuring transaction simulation, threat scanning, MEV protection, and up to $10,000 coverage, aiming to secure AI-driven crypto trades.

Research with immediate practical relevance

No significant new developments.