Agentic AI scales up; global regulation intensifies

Anthropic / Claude ecosystem

Anthropic launches 10 Claude agent templates for financial services

Ready-to-run Claude agent templates for financial services aim to accelerate enterprise AI deployments in banking and insurance.

A context-blind argument parsing vulnerability in Claude Code 2.1.118 could let attackers execute commands, now fixed.

Anthropic CEO: Claude Co-work agent feature developed autonomously by Claude, user growth surges

In an interview, CEO Dario Amodei reveals that a new Co-work feature was largely built by Claude itself, with rapid adoption, and hints at potential free access in the future.

Frontier model providers

Meta opens Ray-Ban Display glasses to developers via two SDK paths

Developers can now build for Meta's display glasses using native toolkits or browser-based web apps for rapid prototyping.

OpenAI launches GPT-5.4-Cyber model to counter Anthropic in cybersecurity race

New model integrates into enterprise workflows for cybersecurity tasks, directly competing with Anthropic's offerings.

Google updates spam policy to ban manipulation of AI-generated search overviews

New policy explicitly prohibits techniques designed to influence AI Overviews and AI Mode in Search, a move to maintain search integrity.

Google Drive adds AI 'Projects' feature for collaborative document management

Project-level AI chatbot centralizes documents and conversation history for enhanced team collaboration.

AI developer tooling & infrastructure

Cursor releases Composer 2.5, its most powerful model for code generation

Composer 2.5 boosts intelligence, long-running task reliability, and overall performance.

Omium launches AI agent observability and reliability platform

New tool aims to bring production-grade monitoring and reliability to AI agent deployments.

Cloud & platform providers

Cloudflare and Stripe enable AI agents to autonomously create accounts, buy domains, and deploy services

Stripe Projects and Cloudflare protocol allow agents to handle payments and provisioning with minimal human intervention.

AI policy, regulation & governance

Pope Leo XIV to issue first AI encyclical 'Magnifica Humanitas' with Anthropic co-founder

On May 25, Pope Leo XIV will launch the Catholic Church's first encyclical focused on artificial intelligence and human dignity, with Anthropic co-founder present, marking a historic Vatican engagement with AI ethics.

Anthropic to brief Financial Stability Board on cyber flaws exposed by Mythos model

Anthropic will present findings on cybersecurity risks revealed by its Mythos model to global financial regulators, highlighting AI's potential to uncover systemic vulnerabilities.

G20 seeks answers on Claude Mythos risks; Anthropic ready to brief leaders

Anthropic prepares to discuss the security implications of its Mythos model with G20 leaders amid concerns over AI-enabled cyber threats.

EU Digital Omnibus on AI bans nudifier apps, strengthens deepfake protections

New provision under the EU's AI legislation explicitly outlaws nudifier applications, tightening controls on AI-generated sexual content.

Bipartisan House bill would require federal agencies to adopt NIST AI risk management framework

The proposed legislation mandates agencies to use the NIST AI RMF and train staff for AI system acquisitions.

China issues first regulations for anthropomorphic AI interaction services

New interim rules require providers of AI systems with human-like interaction to ensure transparency and content safety.

Australia's federal budget includes $70M in grants to advance industry-specific AI beyond chatbots

Funding targets local AI capability and includes instant asset write-offs to encourage adoption.

Jury rejects Elon Musk's $150 billion lawsuit against OpenAI and Sam Altman

California jury dismisses the case, upholding statute-of-limitations defense, removing a major legal threat.

Industry & market moves

Salesforce plans $300 million Anthropic token spend in 2026 for AI coding

CEO Marc Benioff says the investment will boost AI-powered software development across the company.

Commonwealth Bank of Australia appoints Mary-Anne Williams as first chief AI scientist

This is the first such role at a major Australian lender, signaling increased focus on AI leadership in banking.

Multiplayer AI startup Dust raises $40M to build enterprise collaborative AI platform

Dust's platform enables multiple AI assistants to work together across enterprises, moving beyond isolated chatbots.

Decart raises $300M to deploy real-time world model on Amazon chips

Funding from multiple investors will accelerate the AI research lab's real-time video and world model technology.

NEXT Investors backs Transient.AI Series A to bring AI to regulated capital markets

The investment will support Transient.AI's development of AI-native investment management tools for institutional trading.

AI sales compensation startup Dolfin raises €2.1M seed to replace spreadsheets with real-time incentive intelligence

The platform uses AI to automate and optimize sales compensation management.

AI product & feature launches

BUZZ HPC announces 320 MW sovereign AI data center in Greater Toronto Area

Planned AI gigafactory with 100,000 GPUs will be one of Canada's largest, aiming to provide sovereign AI capacity.

Ex-eBay employee builds AI marketing firm operated entirely by 27 autonomous agents with no human employees

The company runs on AI agents that handle all functions from client acquisition to content creation.

Zurich Insurance rolls out agentic AI underwriting with Cytora platform globally

The deployment allows AI to digitize risk submission data across languages, accelerating underwriting processes.

Salesforce showcases deployed enterprise AI agents at Agentforce Demo Day

Real customer stories and demos of AI agents on the Agentforce platform highlight ROI and new deployment patterns.

PolyAI opens its agentic dialog platform to all builders, democratizing complex conversational AI

Previously enterprise-only platform now available for any developer to create production-grade voice and chat agents.

Sapient Intelligence launches HRM-Text, a brain-inspired 1B parameter model trained on 40B tokens

The model claims competitive reasoning with far fewer tokens, challenging the dominance of large LLMs.

Augury launches Industrial AI Workforce with role-based agents for self-optimizing production

Built on Augury Machine Health, AVEVA CONNECT, and Google Gemini, the platform translates machine insights into actionable production actions.

Denodo extends data virtualization for agentic AI with new AWS integrations

Denodo's platform now provides governed, semantic data access for AI agents across hybrid clouds via AWS services.

Research with immediate practical relevance

Researchers reveal MetaBackdoor: AI backdoor attack exploiting input length to evade defenses

The attack vector bypasses content filters and can stealthily trigger data leakage in LLMs.