AI agents drive enterprise adoption; Global regulation…

Anthropic / Claude ecosystem

Anthropic's Code With Claude Announces Managed Agents ... - InfoQ

Anthropic's Code with Claude 2026 conference unveiled significant feature releases, including managed agents, proactive workflows, and capability improvements. The company also announced an 80x annualized revenue growth for Q1 2026, significantly exceeding its planned 10x target.

Frontier model providers

OpenAI updates ChatGPT safety for high-risk conversations

OpenAI has introduced 'safety summaries' for ChatGPT, a new feature designed to help the model recognize and adapt to evolving safety risks in conversations. This update has improved harm-detection performance by 39–52% in high-risk scenarios, enhancing the platform's ability to identify and mitigate potential misuse.

AI developer tooling & infrastructure

A critical Remote Code Execution (RCE) vulnerability was discovered in Anthropic's Claude Code (version 2.1.118), allowing arbitrary command execution through malicious deeplinks. The flaw has since been patched, addressing a significant security risk for developers using the platform.

Cursor releases Composer 2.5 as its most powerful model with gains in intelligence and reliability · Digg

Cursor has released Composer 2.5, its latest and most powerful model, featuring improved intelligence and sustained performance for long-running tasks. This release marks the beginning of joint work with SpaceXAI, indicating a push towards more robust and reliable AI coding assistants.

Chrome DevTools for coding agents · GitHub

Google Chrome DevTools has released MCP v0.26.0, enabling coding agents like Claude and Gemini to control and inspect live Chrome browsers via the Model-Context-Protocol. This allows for more reliable automation and debugging of web-based tasks by AI agents.

Context is the New Bottleneck: Building Token-Efficient AI Coding Agents with MCP in 2026

New research highlights token efficiency as the defining engineering challenge for AI agents in 2026, with the Model Context Protocol (MCP) standardizing tool integration. Semantic retrieval strategies are replacing simple grep-based context gathering to build more token-efficient AI coding agents.

Cloud & platform providers

谷歌雲端硬碟導入AI「專案」功能 助用戶高效管理協作 | 蕃新聞

Google Drive has integrated Gemini AI as a new project management tool, allowing users to create dedicated AI chatbots. These chatbots can be scoped to specific projects, centralizing document and conversation management for enhanced collaboration and efficiency.

Cloudflare and Stripe Let AI Agents Create Accounts, Buy Domains ...

Cloudflare and Stripe have launched the first major cloud provider protocol enabling AI agents to autonomously provision accounts, purchase domains, and deploy to production with minimal human intervention. This is backed by Stripe's identity and payment infrastructure.

AI policy, regulation & governance

Is EU legislation protecting you from AI deepfakes? Ask the Euronews AI chatbot

The EU has banned non-consensual deepfake pornography, specifically 'nudifier apps,' under the Digital Omnibus on AI. The compliance deadline for this new regulation is December 2, 2026, marking a significant step in addressing AI-driven harm.

Federal agencies would use NIST’s AI guidelines under bipartisan House bill

A new bipartisan bill in the United States House of Representatives proposes to mandate federal agencies adopt NIST's AI Risk Management Framework. This legislation would transform voluntary guidelines into enforceable standards for government AI procurement and operations.

China’s Interim Measures for the Administration of Anthropomorphic AI Interaction Services: A New Phase in the Regulation of Human–AI Interaction

China's Cyberspace Administration has released its first regulatory framework specifically targeting AI-driven emotional interaction and anthropomorphic services. These 'Interim Measures' signify a shift towards scenario-based AI regulation with enhanced safeguards for minors and user well-being.

Industry & market moves

FinancialContent - HIVE's BUZZ HPC Announces 320 MW Sovereign AI Infrastructure in Greater Toronto Area

HIVE Digital Technologies and BUZZ High Performance Computing have announced the construction of Canada's largest AI gigafactory in the Greater Toronto Area. The facility will have a 320 MW capacity and over 100,000 GPU capability, aiming for a late 2027 launch as part of a sovereign AI infrastructure strategy.

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

The Commonwealth Bank of Australia has appointed Mary-Anne Williams as its first-ever chief AI scientist. Williams brings frontier research expertise from UNSW to lead the bank's AI innovation strategy, signaling a deeper commitment to advanced AI development.

Multiplayer AI startup Dust raises $40M to help enterprises move beyond isolated AI assistants - SiliconANGLE

Dust (Permutation Labs SAS) has raised a $40M Series B funding round to scale its multiplayer AI platform. The platform enables cross-organizational agent collaboration and shared governance, reaching 3,000 customers with zero churn in 2025.

Roadrunner Raises $27M in Funding

AI-native revenue infrastructure startup Roadrunner has successfully raised $27M across Seed and Series A funding rounds. This capital injection will be used to scale the team and accelerate the development of its quote-to-cash platform.

Decart raises $300M to put a real-time world model in front of Amazon’s chips

Decart has successfully raised $300M in Series funding, bringing its total funding to over $450M. The capital will be used to scale real-time world models across Nvidia, Google, and Amazon chips, signaling a move towards vertically integrated inference optimization.

Searchable secures £10.3m to boost its AI performance marketing platform - UKTN

London-based AI search marketing startup Searchable has secured a £10.3m Series funding round led by Headline. The funding will accelerate product development and support US and UK expansion.

NEXT Investors Invests in Transient.AI Series A to Advance AI Innovation in Regulated Capital Markets | National Business | joplinglobe.com

NEXT Investors has led a Series A financing round for Transient.AI, an AI platform specifically designed for institutional trading in regulated capital markets. This investment aims to advance AI innovation within a highly regulated sector.

AI product & feature launches

Global Big Tech Joins AI Firms in Full-Scale AI Agent Showdown - The Asia Business Daily

Global tech giants like Apple, Google, OpenAI, and Anthropic are entering a full-scale competition to integrate AI agents directly into operating systems and core products. This battle for AI agent dominance is projected to grow the market from $9B to $45B by 2030, with major players releasing new agentic features and platforms.

TeamCentral Launches Central AI and CORBITM to Turn Enterprise AI Agents Into Secure, Governed Business Execution

TeamCentral has launched Central AI, an enterprise AI agent platform designed to unify data across disparate systems and orchestrate AI agents with role-based security and governed execution capabilities. The platform aims to provide secure and compliant business automation through intelligent agents.

South Africa leads rest of Africa in AI adoption - Microsoft

Microsoft's Q1 2026 Global AI Diffusion report indicates South Africa leads AI adoption in Africa, with 23.1% of its working-age population using AI. However, the report also reveals a widening gap in generative AI usage between the Global North (27.5%) and Global South (15.4%).

CSIRO launches new AI infrastructure to help robots and machines learn in real time

CSIRO has launched Vetra AI infrastructure, bringing high-performance AI computing to the edge near robots and sensors. This enables real-time learning in physical environments, with reduced environmental impact through CO2-based cooling.

Zurich expands agentic AI rollout with Cytora partnership

Zurich Insurance has deployed Cytora's agentic AI platform across five countries, significantly reducing manual underwriting triage time by 80% and increasing straight-through processing from 10% to 95%. This represents one of the largest recent deployments of agentic AI in commercial insurance underwriting.

Sapient Intelligence launches HRM-Text, challenging the LLM monopoly with a brain-inspired foundation model trained on up to 1000x fewer tokens

Sapient Intelligence has launched HRM-Text, a brain-inspired foundation model that achieves competitive reasoning performance with up to 1000 times fewer training tokens and a training cost of only $1000. This challenges the prevailing reliance on massive, trillion-parameter systems.

PolyAI opens its Agentic Dialog Platform, making the tech behind complex conversations for hundreds of enterprises available to every builder

PolyAI's enterprise-grade dialog platform, previously exclusive to hundreds of enterprise customers, is now open to all builders, offering free access for the first two months. This platform powers complex conversations for advanced AI agents.

Augury Shaping the Future of Production with the Industrial AI Workforce

Augury has launched its Industrial AI Workforce, a suite of role-based AI agents designed to automate industrial workflows. These agents combine machine health data with Google Gemini reasoning and AVEVA operational context to enhance predictive maintenance and operational efficiency.

Denodo Advances Trusted Data Foundations for Agentic AI

Denodo Technologies Inc. has extended its logical data management platform to provide real-time, governed data access for enterprise AI agents across hybrid and multi-cloud environments. This is achieved through new integrations with AWS services like SageMaker, Bedrock AgentCore, and Quick.

Research with immediate practical relevance

The AI backdoor your security stack is not built to see - Help Net Security

Joint research by Microsoft and the Institute of Science Tokyo has uncovered MetaBackdoor, a novel LLM backdoor attack. This attack uses input length as a hidden trigger, bypassing existing content filters and persisting through fine-tuning, enabling system prompt theft and autonomous data exfiltration.