The Informed Layer · Newsscan
Daily briefing

Anthropic agents advance; Global AI regulation heats up

← All issues

Anthropic / Claude ecosystem

New in Claude Managed Agents: run agents on a schedule and store environment variables in vaults | Claude

Anthropic has enhanced Claude Managed Agents with new capabilities, allowing users to schedule the execution of agentic tasks and securely store environment variables in vaults. This enables autonomous, recurring workflows and secure access to CLI-based tools without exposing API keys, significantly improving enterprise automation possibilities.

Claude Adds Observability Tools for MCP Connectors in Beta - Blockchain.News

Anthropic's Claude has launched a new observability dashboard for its Model Context Protocol (MCP) connectors, now available in public beta. This tool allows developers to monitor key metrics such as adoption rates, error rates, latency, and per-tool health diagnostics for their MCP integrations.

Frontier model providers

OpenAI Responses API Can Now Search the Web for Images

OpenAI's Responses API web_search tool has been updated to return image results alongside text in a single unified call. This new functionality streamlines the process for applications, such as e-commerce or travel platforms, to display visual content without needing to make separate API requests.

Cohere open-sources a coding agent that runs on a single H100 | VentureBeat

Cohere has open-sourced 'North Mini Code,' a 30B Mixture-of-Experts (MoE) coding agent designed to run efficiently on a single NVIDIA H100 GPU. This release aims to provide an accessible local deployment option for enterprises, offering an alternative to proprietary managed coding models.

AI developer tooling & infrastructure

Cursor's 18-Month Data Shows Vibe Coding's Real State: AI Widens the Developer Gap

An 18-month dataset from AI coding assistant Cursor reveals that AI coding is exacerbating, rather than closing, the developer skill gap. The data indicates that the top 1% of users are producing 46 times more code than median users, suggesting AI empowers highly skilled developers disproportionately.

Nexla Launches MCP Studio to Build Governed, Task-Specific MCP Servers Across 600+ Enterprise Systems | Markets Insider

Nexla has launched MCP Studio, a new platform designed to enable enterprises to quickly build governed, task-specific Model Context Protocol (MCP) servers. This studio allows for conversational setup across over 600 integrated systems, bypassing traditional application-by-application data integration constraints.

Cloud & platform providers

Announcing Spark Connect on Amazon EMR Serverless: Interactive PySpark development, anywhere | AWS Big Data Blog

AWS has introduced Spark Connect support for Amazon EMR Serverless, allowing developers to perform interactive PySpark development locally while executing full-scale operations on remote EMR infrastructure. This eliminates the need for repeated deployment cycles, accelerating development workflows.

Route public traffic to private applications with Cloudflare

Cloudflare is now extending its full suite of security, performance, and programmability services to private applications without requiring public IP exposure or connector software. This unifies public and private traffic routing, enhancing protection and efficiency for internal enterprise resources.

AI Traffic Grew 6.5x Faster Than Human Traffic This Year | Fastly

Fastly's report indicates that AI traffic increased by 30% from January to May 2026, growing 6.5 times faster than human traffic. Notably, Claude traffic surged by 555%, and AI requests now require direct origin access 51% of the time, compared to just 9% for human requests.

AI policy, regulation & governance

Meta Strips Privacy Control From Billions — Except Where Regulators Can Stop It

Meta is reportedly removing a privacy control feature for billions of users globally that allowed them to limit data usage from activity off Meta technologies. However, this change will not apply to users in highly regulated markets such as the EU, UK, and Brazil, illustrating a two-tiered privacy approach influenced by regional regulatory enforcement.

EU tells AI companies how they can make their content detectable | Euractiv

The European Union has issued guidance to AI companies on methods to make synthetic content detectable. This initiative aims to help AI providers and deployers comply with the transparency obligations of the AI Act, specifically concerning the marking and labelling of AI-generated materials.

Sound Practices for Responsible Adoption of Artificial Intelligence (AI): Consultation Report - Financial Stability Board

The Financial Stability Board (FSB) has released a consultation report outlining 12 sound practices for the responsible adoption of AI. These practices are designed to guide financial institutions in managing AI-related risks and ensuring financial stability.

Singapore - Facilitating The Safe Adoption Of AI – New Initiatives Unveiled At Committee Of Supply 2026. - Conventus Law

Singapore has announced a comprehensive set of AI governance and adoption initiatives. These include the National AI Impact Programme, targeting 10,000 enterprises and 100,000 workers, along with a Model AI Governance Framework for agentic AI systems and new AI transparency labels.

What APP entities need to know about the APP 1 amendments

New amendments to the Australian Privacy Principles (APPs 1.7–1.9) introduce mandatory disclosure requirements for automated decision-making (ADM) in privacy policies. The Office of the Australian Information Commissioner (OAIC) is expected to provide further guidance by September 2026.

Senator seeks admin leadership to testify on AI policy - Route Fifty

Senator Elizabeth Warren is requesting direct testimony from Trump administration officials regarding AI regulation policy and the associated financial and infrastructure risks. This move signals increasing congressional scrutiny over the government's approach to AI.

Australia's AI Regulation Set for Crucial Review | Mirage News

The Australian Senate has commenced an inquiry into the impacts of AI data centers on energy, water, and local communities. The Senate has opened submissions for this major review, indicating intensifying regulatory scrutiny over AI infrastructure.

Decisions in the dark no more – OAIC soon to turn the lights on for ADM transparency obligation | Herbert Smith Freehills Kramer - JDSupra

The Office of the Australian Information Commissioner (OAIC) is set to release regulatory guidance on automated decision-making (ADM) transparency obligations under APPs 1.7–1.9. This guidance, effective December 10, 2026, will clarify disclosure requirements for organizations using AI and algorithmic systems.

Australia Results Report 2023-2025 – For Public Comment | Open Government Partnership

Australia's third Open Government Partnership (OGP) action plan for 2023-2025 includes new commitments related to a national artificial intelligence strategy, amendments to privacy legislation for automated decision-making, youth advisory groups, and beneficial ownership transparency.

Fully autonomous drones have killed human soldiers for the first time | New Scientist

For the first time, fully autonomous drones, specifically Ukraine's 'Terminator' system, have reportedly killed human soldiers without any human oversight or remote control. This marks a significant and concerning milestone in the deployment of autonomous weapons in active warfare.

Industry & market moves

Can 15 European startups reshape physical AI? Google DeepMind just bet on them

Google DeepMind has launched its inaugural Accelerator: Robotics cohort, selecting 15 European startups focused on physical AI hardware. These startups, ranging from medical microrobots to electronic skin, will receive equity-free mentorship and up to $350,000 in Google Cloud credits.

xAI is reportedly transitioning its Grok training operations from a pure research focus to an infrastructure engineering discipline. The company has hired a veteran Starlink engineer to manage the reliability and scalability of its GPU clusters, signaling a strategic shift towards robust, production-grade AI infrastructure.

Warner Music Group Acquires Sureel AI to Protect Artists' Image, Work

Warner Music Group (WMG) has acquired Sureel AI, a company specializing in protecting artists from unauthorized AI training and monetization of their intellectual property, including voices and likenesses. This strategic acquisition reinforces WMG's commitment to safeguarding artist rights in the age of generative AI.

Figure Enters into Agreement to Acquire Kiavi, Bolstering Blockchain-Native Marketplace | Markets Insider

Figure Technology Solutions, Inc. has agreed to acquire Kiavi, an AI-powered lending platform for residential transition loans (RTL) and debt service coverage ratio (DSCR) loans, for $717 million. This acquisition will add $7 billion in annual first-lien volume to Figure's blockchain-native marketplace.

Jedify Raises $24 Million in Series A Funding to Build Context Graphs for Enterprise AI Agents

Jedify has successfully closed a $24 million Series A funding round to develop autonomous context graphs for enterprise AI agents. This funding aims to address the critical challenge of equipping AI systems with deep, real-time business context at runtime.

Exclusive: Relai raises $6.9M to enable verifiable and continuous learning for AI agents - SiliconANGLE

Relai Inc. has secured $6.9 million in pre-seed and pre-pre-seed funding to develop a verifiable continual learning platform for AI agents. This platform aims to ensure that autonomous agents remain reliable in production as they continuously learn and improve over time.

Poetic emerges from stealth with $50M funding from OpenAI to automate enterprise tasks

Poetic, an OpenAI-backed enterprise automation startup, has emerged from stealth mode with $50 million in Series funding. This investment signifies strong investor confidence in AI-driven workflow automation as a commercially viable application of large language models.

Exclusive: The Cork fintech that started as a Liverpool ticket escrow raises $10M to become the trust layer for AI shopping agents — TFN

Cork-based fintech company, Trustap, has raised $10 million in Series funding, signaling a pivot from its original ticket escrow service. The company now aims to establish itself as the critical trust layer for AI shopping agents, addressing security and reliability in AI-driven e-commerce.

Rivvun AI raises $7.55M to recover enterprise contract leakage

Former Icertis executives have launched Rivvun AI, raising $7.55 million in seed funding for their autonomous AI agents. These agents are designed to recover enterprise spend lost between contract obligations and financial settlement, addressing a significant pain point in corporate finance.

AI product & feature launches

New Relic Introduces AI Coding Observability to Bring Critical Visibility and Governance to AI Coding Assistants | Business News This Week

New Relic has launched AI Coding Observability, extending production-grade monitoring capabilities into the AI coding phase. This new offering provides vendor-neutral visibility across various AI coding assistants, including Claude Code, Cursor, GitHub Copilot, Windsurf, and Amazon Q.

GENISOM AI Debuts at ICRA 2026 with Full-Stack Embodied

GENISOM AI has made its debut at ICRA 2026, showcasing a complete full-stack embodied intelligence ecosystem. This includes industrial-grade quadruped robots (M1, L1 series) with a 1:1 load-to-weight ratio, an open-source MATRiX simulation platform, RoamerX navigation system, and the SomaMind agentic framework.

Decart's new world model can simulate hours of photorealistic driving — with some caveats | TechCrunch

Decart has unveiled Oasis 3, a new world model capable of generating hours of photorealistic driving scenarios with real-time interactivity. The model is available via API at a cost of $0.02 per second.

New Z01 humanoid robot performs complex Tai Chi routine with stability

Zoomlion's new Z01 humanoid robot has demonstrated advanced motion control and dynamic balance by performing a complex Tai Chi routine. This showcases the integration of embodied AI in industrial robotics, indicating significant progress in dexterity and stability.

Exclusive: MotherDuck adds agentic data ingestion to its cloud analytics service - SiliconANGLE

MotherDuck has launched 'Flights,' an agentic data ingestion feature for its cloud analytics service. This new capability automates data pipeline creation and management through natural-language AI interactions, consolidating the analytics stack into a single platform.

Research with immediate practical relevance

HPCwire - Since 1987 – Covering the Fastest Computers in the World and the People Who Run Them

Researchers at Los Alamos National Laboratory have developed a real-time, plug-and-play tool called Prelim Attention Score (PAS) for detecting hallucinations in vision-language models. The tool monitors attention patterns and achieves state-of-the-art accuracy with minimal computational overhead.