The Informed Layer · Newsscan
Daily briefing

Model race intensifies; regulation and M&A surge

← All issues

Anthropic / Claude ecosystem

Anthropic Launches Official Claude Code Plugin Directory with Security Vetting

Anthropic has formalized Claude Code plugins with an official, vetted directory and a security/screening process, including an 'Anthropic Verified' badge. This move aims to mitigate risks from unverified MCP connections.

Frontier model providers

Google DeepMind Fuses Street View with Genie AI for Playable Real-World Sims

Google DeepMind integrated Street View imagery with its Genie AI model to create interactive, modifiable simulations of real-world locations, announced at Google I/O 2026.

Google CEO Admits Coding Gap, Plans to Close It with Gemini 3.5 Flash

Google CEO acknowledged a coding capability gap in the AI race, detailing plans to deploy faster, cheaper Gemini 3.5 Flash models across its ecosystem to catch up.

xAI Completes Training of Grok V9 Medium with 1.5T Parameters

Elon Musk confirmed that xAI has finished training Grok V9 Medium, a 1.5-trillion-parameter model aimed at fixing data quality issues and boosting coding performance, with public rollout in 2-3 weeks.

Mistral AI Releases 8x22B MoE Model with 65k Context Window

Mistral AI published a new 8x22B Mixture-of-Experts model featuring 56 layers, 48 attention heads, and a 65k token context window, offering a competitive open-weight option.

Google CodeMender Agent Autonomously Patches Code Vulnerabilities

Google introduced CodeMender, an AI agent that autonomously patches security vulnerabilities end-to-end and upstreams fixes to open-source projects, with human-in-the-loop approval.

Critical Gemini Vulnerabilities Fixed; 93% of Integrations Must Upgrade Within 48 Hours

Three high-severity CVEs were disclosed for Gemini's model inference service, requiring immediate upgrades for high-concurrency scenarios, alongside an official rollback plan.

AI developer tooling & infrastructure

No significant new developments.

Cloud & platform providers

AWS Expands P5.48xl GPU Instances for SageMaker Studio Notebooks

AWS generally available the EC2 P5.48xl instances in new regions for SageMaker Studio notebooks, providing high-performance GPU compute for AI workloads.

AI policy, regulation & governance

xAI Adds $2.8B in Gas Turbines Amid Clean Air Act Lawsuit

xAI is investing $2.8 billion in natural gas turbines for its Colossus 2 AI compute site, even as the NAACP sues under the Clean Air Act, while SpaceX pitches orbital solar as a long-term alternative.

UL Tightens AI Failure Testing for CNC Systems, Effective May 22, 2026

UL Solutions issued a bulletin mandating AI failure safety validation for CNC controls, expanding safety compliance requirements for the North American market.

Massachusetts Proposes Ban on Weaponized Robots and Drones

A new bill in Massachusetts seeks to prohibit the manufacturing, modification, sale, possession, or operation of weaponized robots, signalling regulatory movement on robotics safety.

Australian media organisations and industry bodies issued a joint statement urging the federal government to maintain strict copyright safeguards amid AI developments.

Australia and the UK Sign AI Safety and Security Cooperation MoU

Australia and the United Kingdom signed a memorandum of understanding to jointly advance AI safety, security, and governance, the first formal bilateral agreement between the two nations on AI.

Australian Employer Group Demands Crackdown on AI-Generated Fair Work Claims

The Australian Resources and Energy Employer Association urged changes to Fair Work Commission processes to handle a flood of AI-generated, potentially frivolous claims.

Industry & market moves

Meta Cuts 15,000 Jobs in AI-First Restructuring

Meta is reportedly slashing 15,000 positions as part of a large-scale restructuring to prioritize AI, a higher number than previously reported layoffs.

Cohere Acquires Reliant AI to Boost Pharma AI Platform

Cohere expanded its presence in Germany by acquiring Berlin-based Reliant AI, a move to enhance its North platform for pharmaceutical applications.

Intuit Cuts 3,000 Jobs, Then Integrates TurboTax into Claude and ChatGPT

Intuit executed mass layoffs while simultaneously signing deals with Anthropic and OpenAI to embed TurboTax, QuickBooks, Credit Karma, and Mailchimp inside Claude and ChatGPT.

OpenAI Opens First Applied AI Lab Outside the US in Singapore

OpenAI established a Singapore-based Applied AI Lab through a new partnership, marking its first such facility outside the US as IMDA updates its AI governance framework.

Samsung SDS and KKR Launch Startech AI to Formalize Global Infrastructure Alliance

Samsung SDS and private equity firm KKR formed a New York-based entity, Startech AI, to institutionalize an ongoing partnership for AI infrastructure investments and M&A.

SAP to Acquire Dremio and Prior Labs to Lead Enterprise Agentic AI

SAP announced plans to purchase data platform Dremio and AI startup Prior Labs to bolster its agentic and tabular foundation model strategy.

Expert.ai and Fincons Group Partner for Neuro-Symbolic AI in Regulated Industries

Expert.ai and Fincons Group formed a partnership to bring neuro-symbolic AI to data-driven businesses, enabling transparent and robust AI for large-scale, regulated deployments.

Outcomes4Me Acquires Geno.Me and Realyze Intelligence to Expand AI Cancer Care

AI-driven cancer care platform Outcomes4Me acquired genomic data and patient-matching companies Geno.Me and Realyze Intelligence to broaden clinical trial access.

NEURA Robotics Partners with Bosch to Advance German-Made AI Robotics

NEURA Robotics and Bosch partnered to fuse real-world data with AI via NEURA's Neuraverse ecosystem, aiming to accelerate deployment of industrial and service robots like the Quadruped 4NE1.

Qualtrics Acquires Press Ganey Forsta for $6.75B, Expanding AI Healthcare Data

Qualtrics acquired experience management firm Press Ganey Forsta to combine data from over 41,000 healthcare facilities, enabling predictive, AI-driven decision-making in healthcare.

AI product & feature launches

SAP Unveils Sovereign EU AI Cloud with Data Residency

SAP launched a European sovereign AI and cloud offering that includes data residency and on-site deployment options, targeting enterprises with strict compliance needs.

LSEG's MCP Server Seen as AI Data Distribution Turning Point

A leading investment bank views LSEG's release of a Model Context Protocol server as evidence that AI can become a distribution channel for financial data, potentially generating incremental revenue.

Fujitsu Develops Self-Evolving Multi-Agent AI for Business Operations

Fujitsu introduced autonomous, self-evolving AI agents that learn from results, human feedback, and business rule changes to continuously improve operational workflows.

Workday Launches AI Agents for IT Support and Travel Management

Workday introduced Sana for IT Service Management and Sana Travel Agent, AI-powered agents that unify tasks and expense workflows within the Workday platform.

Research with immediate practical relevance

Embodied AI Qumus Fabricates Graphene Transistor Autonomously

Researchers demonstrated Qumus, an embodied AI system that autonomously fabricated graphene nanodevices and a graphene transistor with minimal human help, published in a scientific paper.

Cerebras Achieves 981 Tokens/sec on 1T-Parameter Kimi K2.6, Claiming 6.7x Speedup

Cerebras Systems ran the 1-trillion-parameter Kimi K2.6 MoE model on its wafer-scale engine, achieving 981 tokens/sec and a 6.7x speedup over the next fastest GPU cloud, validated by a third party.

Frontier Supercomputer Trains AI to Model Cosmic Magnetic Turbulence

Researchers used the Department of Energy's Frontier supercomputer to train an AI model that captures detailed magnetic turbulence in plasma, advancing astrophysics simulations.

CMR-CLIP AI Reads Cardiac MRI Without Labeled Data, Achieving High Accuracy

Carnegie Mellon and Cleveland Clinic developed CMR-CLIP, an AI system that interprets cardiac MRI scans without manually labeled training data by aligning image sequences with radiology reports.