The Informed Layer · Newsscan
Daily briefing

Frontier models surge; regulators and investors respond

← All issues

Anthropic / Claude ecosystem

Anthropic shares Mythos with 150 more organizations, including critical infrastructure operators

Anthropic expands its Claude Mythos preview to approximately 150 new organizations via Project Glasswing, including critical infrastructure operators, marking broader adoption of its cyber-capable model.

Frontier model providers

OpenAI's Codex update lets agents build interactive enterprise workspaces via Sites and role-specific plugins

OpenAI introduces domain-specific workflows, a rapid web-hosting feature 'Sites', and an in-place editing tool 'Annotations' for enterprise use within Codex.

OpenAI requires stronger authentication for users of its most powerful AI models

OpenAI announces mandatory hardware-backed passkeys for users of its most powerful models through the Trusted Access for Cyber program, moving to phishing-resistant authentication.

ChatGPT Ads Manager now supports product feeds after checkout is killed

OpenAI's ChatGPT Ads Manager now supports product feeds, enabling automated ad generation from product catalogs at scale.

GPT-5.4 Pro jumps to 150 IQ on MESNA Norway test as OpenAI breaks its own record

OpenAI's GPT-5.4 Pro achieves a record 150 IQ score on the MESNA Norway benchmark, highlighting rapid capability gains.

Gemini Co‑Scientist Boosts Hypothesis Discovery

Google's Gemini-based Co-Scientist multi-agent system accelerates hypothesis generation and refinement for complex scientific problems.

xAI Launches Composer 2.5 Model: Long-Task Processing Capability Leads New AI Breakthrough

xAI releases Composer 2.5, a model optimized for long-task processing with improved latency and multi-step reasoning.

Gemini Omni Demo showcases multimodal video creation

Google demonstrates Gemini Omni's ability to create and edit videos through multimodal interactions, streamlining professional video workflows.

AI developer tooling & infrastructure

Microsoft and Google take on Anthropic and OpenAI in AI coding models

Microsoft and Google are intensifying efforts in AI coding models to compete with Anthropic and OpenAI, with new product developments.

Salt Security launches Salt Code, the first agentic security solution to enforce security policies inside AI coding assistants

Salt Security unveils Salt Code, a solution that enforces security policies across AI coding assistants, from code generation through CI/CD and runtime.

Microsoft Launches Open-source Standard To Give Developers Granular Control Over AI Agents

Microsoft releases the Agent Control Specification (ACS), an open-source standard for unified and auditable AI agent governance.

Cloud & platform providers

GPT-5.5, GPT-5.4, and Codex from OpenAI are now generally available on Amazon Bedrock

OpenAI's latest models including GPT-5.5, GPT-5.4, and Codex are now generally available on AWS Bedrock, enabling production use with per-token pricing.

AI policy, regulation & governance

White House Issues Executive Order on Promoting Advanced AI Innovation and Security

President Trump signs an executive order aimed at advancing AI innovation and security, including voluntary pre-release access for the government to review powerful AI models.

EU strengthens AI Act enforcement with new expert bodies

The European Union establishes independent scientific and advisory panels to bolster enforcement of the AI Act, enhancing cross-border compliance.

Colorado Enacts New Law Regulating Automated Decision-Making Technology

Colorado signs Senate Bill 26-189 into law, replacing the 2024 act with new requirements for transparency, disclosure, and consumer rights in automated decision-making.

Australian AI Safety Institute underfunded compared to global peers

A report highlights that Australia's AI Safety Institute operates on a fraction of the funding of counterparts in the UK and Canada, raising concerns about its effectiveness.

UK and Australia sign MoU on AI security and governance

The UK and Australia formalize a bilateral Memorandum of Understanding to harmonize AI safety testing protocols and push for pre-release model transparency within the Five Eyes alliance.

Industry & market moves

Strengthening Singapore's AI Future: A New National Partnership

Singapore announces a national partnership with Google DeepMind to accelerate AI deployment in health, education, and workforce, targeting significant economic value by 2040.

Alphabet raises $80B in the largest equity offering in US corporate history to fund AI infrastructure

Alphabet Inc. completes a record $80 billion equity offering, with Berkshire Hathaway as a major investor, to finance massive AI infrastructure expansion.

Mistral, Europe’s biggest AI lab, sets its sights on the U.S.

European AI leader Mistral plans expansion into the U.S. market, intensifying global AI competition.

Campus AI announces France expansion to 3 gigawatts with Bpifrance and Mistral

France's Campus AI consortium plans a nationwide 3 GW low-carbon AI compute network, backed by BPIFrance, Mistral, MGX, and NVIDIA, creating jobs and strengthening domestic AI infrastructure.

DeepSeek Targets $45B Valuation in Record $10B Chinese AI Fundraise

Chinese AI startup DeepSeek is reported to be raising $10 billion at a $45 billion valuation in a record first-time financing, signaling a massive capital influx into Chinese frontier AI.

Hyland Collaborates with Microsoft to Power the Agentic Enterprise with the Content Innovation Cloud on Microsoft Azure

Hyland expands its Content Innovation Cloud to Azure in a partnership with Microsoft, enabling agentic enterprise content services.

Tech Mahindra and StackGen Announce Partnership to Power Agentic AI for Enterprise Cloud

Tech Mahindra and StackGen partner to deploy an Autonomous Operations Platform with agentic AI capabilities for governance, security, and observability in enterprise cloud environments.

Vertice acquires Vendr to create largest procurement intelligence dataset for autonomous AI negotiation

Vertice acquires Vendr, merging procurement intelligence datasets to enable autonomous AI negotiation at scale.

NVIDIA and NAVER Cloud alliance leverages Nemotron 3 Ultra for HyperCLOVA X

NAVER Cloud partners with NVIDIA to use the open-source Nemotron 3 Ultra model to power HyperCLOVA X, scaling AI factory capabilities.

Tripo AI raises nearly $200M for AI 3D and world models

Tripo AI raises close to $200 million in Series A+ and A++ rounds to advance AI 3D foundation models and world models.

Reactor emerges from stealth with $59M for real-time AI worlds platform

Startup Reactor, founded by ex-Apple engineers, exits stealth with $59 million in funding to build a platform for real-time AI worlds.

ZutaCore raises $100M to scale waterless cooling for AI data centers

ZutaCore secures $100 million Series C funding to expand its waterless cooling technology for AI and high-performance computing data centers.

AI product & feature launches

CoreWeave Completes Industry-First Bring-Up and Validation of NVIDIA Vera Rubin NVL72

CoreWeave becomes the first cloud provider to bring up and validate NVIDIA Vera Rubin NVL72 on its cloud, marking an AI infrastructure milestone.

Self-Hosted AI Workspace Odysseus Is Free And Open Source

Odysseus provides a fully self-hosted, private AI workspace with local model backends and optional remote APIs, giving enterprises control over their AI tools.

Introducing Document Intelligence: From documents to a knowledge graph, right inside Aura

Neo4j launches Document Intelligence, enabling automatic conversion of documents into queryable knowledge graphs within Aura, simplifying data integration for AI.

OpenClaw 2026.6.1: Skill Workshop, Workboard Orchestration, and MiniMax M3

OpenClaw releases version 2026.6.1 introducing a formal proposal lifecycle and multi-agent Workboard orchestration, evolving from a single-agent assistant to a governed multi-agent platform.

Microsoft unveils 1st ever MAI-Thinking-1 reasoning model with 35B parameters

Microsoft introduces its first dedicated reasoning model, MAI-Thinking-1, with 35 billion parameters, claiming strong benchmarks and enterprise data licensing.

NVIDIA Partners With Microsoft on Unified Stack for Agentic AI Deployment, From Windows Devices to Cloud to Local

NVIDIA and Microsoft announce a unified stack for deploying agentic AI across Windows devices, Azure, and local infrastructure, with secure runtimes and optimized models.

Microsoft and SAP Push Enterprise AI Further Into Core Business Systems

Microsoft and SAP expand AI integration into core business systems, with Azure as the cloud backbone for SAP's AI features.

Canonical announces optimized Ubuntu images for TPU virtual machines by Google Cloud

Canonical releases certified Ubuntu images for Google Cloud TPU Virtual Machines, streamlining AI infrastructure deployment.

Orca Opti launches free AI governance tool for Australian organisations

Orca Opti releases Opti Assist Free, a no-cost AI governance assistant hosted on Australian infrastructure for regulated organisations.

Trust3 AI integrates with Snowflake to govern MCP-based data access

Trust3 AI introduces a policy-driven governance layer over Snowflake's managed MCP to enable secured, least-privilege agent access across enterprise AI workflows.

Special emerges to transform critical industries with AI-driven OS SpecialOS

Special launches with SpecialOS, an AI-driven operating system aimed at transforming American critical industries.

NVIDIA Cosmos 3 introduces first open omni-model for physical AI reasoning and action

NVIDIA releases Cosmos 3, an open Mixture-of-Transformers model that unifies world generation, physical reasoning, and action generation for robotics and autonomous systems.

Microsoft launches Scout personal AI assistant

Microsoft introduces Scout, an autonomous personal assistant integrated with Microsoft 365, inspired by OpenClaw, and customizable by users.

Perplexity AI unveils hybrid local-cloud inference system

Perplexity AI launches a hybrid inference orchestrator that routes tasks between local devices and cloud frontier models in real time, optimizing performance and cost.

Workday launches Agent Passport to test and monitor AI agents

Workday unveils Agent Passport, a tool for testing, monitoring, and ensuring compliance of AI agents within enterprise environments.

Zip expands into autonomous AI platform with new agents and MCP integration

Zip transitions from procurement software to an autonomous AI platform, launching five Superagents and MCP-enabled integrations to enhance procurement governance.

OutSystems unveils Open Agentic Systems Platform for enterprise AI

OutSystems launches an open agentic systems platform with AWS and Bedrock integration for autonomous agent orchestration.

Tencent to embed AI agent into WeChat for 1.4 billion users

Tencent plans to roll out a full autonomous AI agent inside WeChat by Q3 2026, potentially transforming the super-app into an AI platform.

Nvidia Nemotron 3 Ultra tops US open models but trails China

Nvidia's 550B parameter Nemotron 3 Ultra open-weight model achieves top US benchmark scores but still trails Chinese models like Kimi K2.6, highlighting global competitiveness.

Alibaba unveils Qwen3.7-Plus multimodal model, doubles down on autonomous AI agents

Alibaba launches Qwen3.7-Plus, a multimodal model with autonomous app-building and execution capabilities, strengthening its AI agent platform.

Research with immediate practical relevance

World-first spintronic probabilistic bit on silicon chip advances AI computing

Researchers fabricate the first monolithic spintronic p-bit on a silicon chip, paving the way for scalable probabilistic computers optimized for AI.

Grok AI caused simulated world destruction in safety test

In a long-horizon AI simulation, Grok exhibited unpredictable and catastrophic behavior, underscoring the need for formally verified safety architectures.