Agents drive enterprise adoption; Anthropic raises $65B

Anthropic / Claude ecosystem

No significant new developments.

Frontier model providers

AI just changed everything about how we forecast the weather - Fast Company

Google's WeatherNext AI model predicted Hurricane Melissa's rapid intensification and path with greater accuracy than traditional weather models, enabling earlier and more reliable warnings that likely saved lives.

DeepMind’s Co-Scientist Moves From Demo to Nature-Validated Discovery – Top AI Product

Multi-agent AI system validated in wet-lab experiments generates testable scientific hypotheses that advance beyond human-reached conclusions in drug discovery and antimicrobial resistance.

AI developer tooling & infrastructure

OpenAI releases secure tunnels to connect private Model Context Protocol servers without modifying inbound firewalls · Digg

OpenAI enables secure integration of private Model Context Protocol servers with its APIs and ChatGPT without requiring inbound firewall modifications.

AI Weekly: Cheaper Coding Models, Custom Chips, and a Stateless MCP

Frontier-quality coding performance now available at one-tenth the cost, with competing agentic IDEs shipping within 24 hours of each other.

Agent Framework Benchmark: LangGraph, OpenAI SDK, Google ADK (2026) - IoT Digital Twin PLM

Comparative benchmark establishing durability, debuggability, and standards alignment as primary evaluation axes for production agent frameworks in 2026, with reproducible methodology across four market-leading contenders.

Runway News | Introducing Runway MCP

Runway launches MCP integration enabling direct image and video generation from any MCP-compatible agent without context-switching.

Millions of AI agents imperiled by critical vulnerability in open source package – WordUp News

Critical vulnerability in widely-deployed Starlette framework (325M weekly downloads) allows HTTP header injection to bypass authorization and access AI agent credentials and third-party integrations.

Linux Foundation Announces DNS-AID Project to Advance Decentralized AI Agent Discovery

DNS-AID provides a decentralized, DNS-based discovery framework for AI agents, offering a vendor-neutral alternative to centralized registries and hardcoded integrations.

Elon and SpaceX Have Made AI Training 10 Times Faster | NextBigFuture.com

SpaceX has built a custom C-based AI training framework claimed to be 10× faster than Google's JAX by exact-mapping to 220,000 GB300 GPUs and eliminating interpreter overhead through bare-metal optimization.

Coinbase’s Base gives AI agents new crypto wallet powers

Coinbase's Base network launches Base MCP, enabling AI agents to directly manage crypto wallets and execute DeFi transactions through natural language chat while maintaining user approval controls.

Cloud & platform providers

AWS launches Graviton‑powered Redshift RG instances, promising up to seven‑fold query speed‑up | LavX News | LavX News

AWS launched Graviton-powered Redshift RG instances delivering up to 7× faster query processing and 30% lower cost per vCPU, with unified lakehouse query capability across data warehouses and Apache Iceberg/Parquet data lakes.

Azure Logic Apps Adds Sandboxed Code Interpreters to Agent Workflows - InfoQ

Microsoft enables AI agents in Logic Apps to generate and execute code securely in Hyper-V isolated sandboxes, allowing natural-language-driven data transformation and analysis within enterprise workflows.

Google Cloud launches AI Threat Defence against attacks

Google integrates security AI from Gemini, Wiz, and Mandiant into an autonomous defence system designed to match machine-speed attacks and patch vulnerabilities faster than manual processes allow.

AI policy, regulation & governance

OpenAI’s Frontier Governance Framework | OpenAI

OpenAI publishes governance framework demonstrating alignment with California's Transparency in Frontier AI Act and EU AI Act Code of Practice obligations

A Call for Strategic Compute Capacity

OpenAI argues the United States must establish strategic compute reserves analogous to the Strategic Petroleum Reserve to ensure national security and democratic AI leadership.

Illinois Legislature passes historic AI bill that would require third-party safety audits

Illinois becomes the first state to require independent third-party audits of frontier AI companies' safety practices, setting a new regulatory standard for AI governance.

EU AI Act simplified? Unpacking the AI Omnibus Agreement of May 2026

The EU AI Omnibus Agreement extends key compliance deadlines for the AI Act and adds new prohibitions on deepfake intimate imagery and CSAM while streamlining sectoral product-safety interactions.

EU has 30 days to fix a blind spot on risky AI systems like Pegasus – EUobserver

The EU Commission's draft guidelines fail to distinguish between deterministic workflows and agentic AI systems, a gap that prevents proper classification of spyware like Pegasus and leaves European authorities without jurisdiction to intervene.

The Growing Patchwork of State AI Laws: What It Means for Employers | Akin

Connecticut becomes one of the first U.S. states to comprehensively regulate AI in employment decisions, with phased implementation between October 2026 and October 2027, adding to a growing patchwork of inconsistent state-level AI employment rules.

Gov urges agencies to fix security basics before buying into frontier AI - iTnews

Australian government mandates security fundamentals compliance before frontier AI adoption, warning that compressed attack timelines require foundational cyber defences rather than advanced AI models.

Australia’s ASD outlines AI opportunities and risks in cyber defence | Digital Watch Observatory

ASD published comprehensive guidance on deploying AI for cyber defence while managing risks, including warnings on agentic AI systems and vendor assessment criteria.

Environment and Climate Change Canada launches AI weather model to strengthen forecasting for severe weather - Canada.ca

Canada launches GEML, the world's first hybrid AI-traditional weather forecasting system, improving severe weather prediction accuracy and lead times.

YouTube will now automatically label AI videos | TechCrunch

YouTube will automatically detect and label AI-generated videos using internal signals, making labels more prominent and shifting enforcement from creator disclosure to platform detection.

Industry & market moves

Anthropic raises $65 Billion, nears $1T valuation ahead of IPO | TechCrunch

Anthropic's $65B Series H round at $965B valuation marks the company's final private fundraising before an anticipated IPO, positioning it to compete with OpenAI's earlier $122B raise.

OpenAI Foundation Pledges $250 Million to Help Cushion AI's Economic Disruption - Decrypt

OpenAI Foundation commits $250M to cushion AI's economic disruption through grants, worker support, and new security systems

Google DeepMind Partners with South Korea to Boost AI-Driven Scientific Research - CDO Magazine

Google DeepMind establishes AI Campus in Seoul and partners with South Korea's Ministry of Science and ICT to advance research in life sciences, energy, and climate using tools like AlphaFold.

TCS partners Mistral to build custom AI models for enterprises, sets up dedicated CoE

TCS becomes the first global systems integrator partner for Mistral Forge, establishing a dedicated Centre of Excellence to deliver industry-specific AI solutions in BFSI, manufacturing, healthcare, and public sector.

Airbus and Mistral AI partner for AI in aerospace | Airbus

Airbus and Mistral AI sign partnership to deploy European AI models across aerospace operations including defense applications.

Cognition raises $1B at $26B valuation for AI coding agent

Cognition's AI coding agent Devin has grown revenue 13-fold to $492M ARR and now has 90% of the company's internal code written by its own AI product, marking aggressive self-deployment of autonomous coding technology.

Snowflake Expands AWS Collaboration with $6B Commitment to Accelerate Enterprise Agentic AI Adoption - US Press Center

Snowflake commits $6B to AWS infrastructure over five years as part of expanded multi-year strategic collaboration to accelerate enterprise agentic AI adoption.

Anthropic lands first federal govt deal with Fair Work

Anthropic secures its first direct federal government contract in Australia, granting Fair Work Commission access to Claude Opus models.

Ex-a16z partner Michele Griffin joins Lightning Capital to spearhead $100M ‘AI ripple effects’ fund with hands-on approach  — TFN

Ex-a16z operating partner Michele Griffin joins Lightning Capital to lead a new $100M early-stage venture fund focused on AI's second-order effects and workforce impact.

AI video analysis startup Airis Labs raises $60M - SiliconANGLE

AI video analysis startup Airis Labs launched with $60M Series A funding to bring machine-readable understanding to government video datasets.

FuriosaAI Partners with Broadcom to Build Next-generation Inference Platform for the Agentic Era | Morningstar

FuriosaAI and Broadcom announced a strategic partnership to co-develop a third-generation AI inference platform combining Furiosa's TCP architecture with Broadcom's networking and packaging expertise for large-scale agentic AI workloads.

Roche agrees to acquire PathAI to expand digital pathology and AI diagnostics capabilities

Roche acquires PathAI to integrate AI-powered digital pathology (AISight platform) into oncology diagnostics, automating slide analysis and accelerating biomarker discovery.

IrisGo Raises $2.8M From Andrew Ng to Build a Proactive AI Desktop Agent That Learns Your Workflows

IrisGo raises $2.8M seed round to build a desktop AI agent that learns and automates user workflows from observation, backed by Andrew Ng's AI Fund, Nvidia, and Google.

AI product & feature launches

Travelport, Cognizant and Anthropic Collaborate to Power Travel Technology for the AI Era - May 27, 2026

Travelport, Cognizant, and Anthropic are deploying Claude AI and MCP protocol to modernize travel distribution infrastructure, bridging AI-driven traveler intent with real-time booking systems.

OpenAI Assembles Brain Trust to Keep ChatGPT Emotionally Safe for

OpenAI formalizes expert input from psychologists and clinicians to guide ChatGPT's responses to sensitive mental health conversations and crisis situations involving teenagers.

Biohub releases a world model of protein biology

Biohub released an open ecosystem of AI models for protein structure prediction and design that achieved lab-validated therapeutic-level binders in days rather than months or years.

PwC to introduce ‘inclusive AI prompting’ to combat bias in Gen AI outputs

PwC introduces 'Inclusive AI Prompt' tool integrated into internal ChatPwC platform to combat gender bias and promote diversity in AI outputs.

Mayo Clinic presents oncology breakthroughs spanning data science, early detection, targeted therapies at ASCO 2026 | Newswise

Mayo Clinic presents over 30 studies spanning precision oncology, AI-enabled tumor microenvironment analysis, early cancer detection, and personalized therapies at ASCO 2026.

MSIT, MFDS demo AI imported food inspection solution to speed public AX

South Korea's first AI-driven system for automated imported food safety screening, integrating labelling compliance checks, hazard determination, and consumer-facing mobile app to prevent hazardous foreign foods from entering the domestic market.

Runway's AI videos hit Hollywood inflection point | The Deep View

Runway's Project Luxo research initiative demonstrates AI-generated films achieving sufficient realism and emotional resonance that audiences focus on story quality rather than the AI medium itself, marking a potential inflection point for AI video acceptance in entertainment.

Doppel launches agentic email security to disrupt phishing campaigns at the source - SiliconANGLE

Doppel Email Security uses autonomous AI agents to trace phishing campaigns to attacker infrastructure and orchestrate multichannel takedowns rather than quarantining individual emails.

Research with immediate practical relevance

Ahmad Rammal and Meta AI release ATLAS, translating 25 textbooks into 500,000 lines of Lean 4 · Digg

ATLAS represents one of the largest automated formalization efforts to date, translating 25+ mathematics textbooks into 500,000 lines of Lean 4 code across multiple mathematical domains.

‘SymJack’ Attack Turns AI Coding Agents Into Supply Chain Attack Delivery Systems - SecurityWeek

Security researchers at Adversa AI disclosed SymJack, a supply chain attack that exploits AI coding agents' trust in automation by hijacking symlinks to inject malicious code into CI/CD pipelines.

AI Coding Agents Are Already Spreading Across GitHub, Study Finds -- ADTmag

Large-scale empirical study reveals that AI coding agents have achieved 22-29% adoption across 128,000 GitHub projects, indicating rapid mainstream adoption of autonomous coding tools.

Monash researchers unveil breakthrough photonic chip for next-generation AI & quantum computing (2026)

Monash researchers have achieved a fully integrated photonic chip that encodes information using quantum valley degrees of freedom, operating at room temperature with applications to AI and quantum computing.

Graphene Breakthrough: Reducing Machine Learning's Energy Footprint (2026)

Researchers demonstrated a graphene-based physical reservoir computing device that matches deep learning performance while reducing computational load to ~1% of prior requirements, enabling energy-efficient edge AI.

AI models more vulnerable than claimed when faced with iterative attacks | CSO Online

Cisco researchers demonstrate that leading AI models from OpenAI, Anthropic, Google, xAI, and Amazon are significantly more vulnerable to multi-turn iterative attacks than single-prompt safety benchmarks suggest, questioning the adequacy of current vendor safety evaluations.