The Informed Layer · Newsscan
Daily briefing

Anthropic extends agentic reach; infra funding swells

← All issues

Anthropic / Claude ecosystem

Anthropic launches Claude Science, an AI workbench for scientists

Anthropic introduced Claude Science, an integrated AI workbench that combines research tools, databases, and Claude models into an auditable, publication-ready workspace. It targets bioscience and life sciences with pre-configured toolkits and scalability. The launch shifts Anthropic's strategy toward vertical workflow platforms for scientific research.

Anthropic releases Claude Sonnet 5, its most agentic model yet

Claude Sonnet 5 is now available with advanced planning, tool use, and autonomous operation capabilities. It offers broad availability and new introductory pricing. The model integrates with cloud services and supports enterprise usage.

Claude in Microsoft Foundry is now generally available

Microsoft announced the general availability of Anthropic Claude models in Microsoft Foundry on Azure. The service enables production-grade agentic AI with enterprise controls, integrating with Azure's governance and compliance tools.

Claude Apps Gateway Debuts for Amazon Bedrock, Google Cloud

Anthropic introduced a self-hosted Claude Apps Gateway that provides centralized controls and cost tracking for Claude Code usage across Amazon Bedrock and Google Cloud. It enables enterprises to manage AI coding agent deployments securely.

Claude Code Attack Uses DNS TXT Payload to Compromise Developer Machines

A newly disclosed attack vector uses DNS TXT records to deliver malicious payloads to Claude Code installations, achieving full system compromise without altering repositories. The vulnerability highlights supply-chain risks in AI developer tools.

Frontier model providers

OpenAI Expands Into Developer Hardware With Codex Micro Keyboard

OpenAI announced Codex Micro, its first branded hardware product developed with Work Louder. The programmable keyboard accessory is designed to accelerate coding workflows for Codex users and is slated for release on July 15, signaling a new hardware direction for the company.

France’s CEA and Mistral AI renew partnership to deploy generative AI platform Maïa

CEA and Mistral AI are strengthening their collaboration to roll out the Maïa generative AI platform across CEA facilities, targeting 6,500+ staff by end of 2026. The partnership focuses on data security and skill-building in French research institutions.

DeepSeek to launch V4 in mid-July with new peak-time API pricing

DeepSeek plans to release its V4 model in mid-July, featuring a 1-million-token context window and a peak/off-peak API pricing structure. The release includes open-source inference acceleration technologies (DSpark/DeepSpec) and dynamic pricing to manage demand.

AI developer tooling & infrastructure

Cursor launches an iPhone app for its AI coding agent

Anysphere released a beta iPhone app for Cursor, its AI coding agent, allowing developers to oversee cloud-based coding tasks and receive Live Activities updates on mobile. The app extends Cursor's reach beyond desktop environments.

X (formerly Twitter) offers a hosted MCP server for AI tool integration

X launched a hosted MCP server that allows AI tools to connect directly to its API using user account permissions. This move simplifies integration for developers and positions X as a real-time data source for AI agents and assistants.

Elastic open-sources Atlas Agent Memory based on cognitive science

Elastic released Atlas, an open-source memory system for AI agents inspired by cognitive science. It structures memories into episodic, semantic, and procedural types with hybrid search, giving agents long-term contextual awareness.

Perforce launches Agentic Gateway to govern AI agents and cut token costs

Perforce released the Agentic Gateway, an orchestration layer that manages AI agent interactions and reduces token costs. Integrated with Perforce's MCP portfolio, it provides enterprises with governance controls for multi-agent systems.

Cloud & platform providers

AWS invests $1 billion in forward-deployed AI engineers

Amazon Web Services is creating a new Forward Deployed Engineering (FDE) organization with $1 billion funding. The FDE team will embed engineers within client companies to accelerate agentic AI solutions and transfer AI skills, aiming for long-term customer self-sufficiency.

AI policy, regulation & governance

Warner bill would create federally vetted list for secure, trustworthy AI agents

A draft Senate bill, the AI AGENT Act, proposes a federally vetted list of AI agent providers meeting security and identity standards enforced by the FTC. The legislation aims to build trust in AI agents for sensitive enterprise and government use.

Vietnam: Law on Digital Transformation enters into force, regulating unilateral conduct

Vietnam's Law on Digital Transformation, effective July 1, includes rules against unilateral conduct abuses, imposing obligations on digital platforms. The law addresses AI-driven digital practices and aims to prevent dominant-position misuse.

Australian audit highlights governance gaps in public-sector AI

An Australian National Audit Office report identified governance weaknesses in the use of AI within a public-sector patent process. It called for stronger cybersecurity governance and clearer monitoring as agencies shift from experimentation to routine AI use.

Industry & market moves

HiddenLayer and Cohere collaborate to remove security barriers to enterprise agentic AI

HiddenLayer and Cohere announced a partnership to integrate security into agentic AI deployments, combining HiddenLayer's AI Security Platform with Cohere's North. The collaboration aims to enable safe, enterprise-scale AI agent operations across business systems.

Rebellions Acquires AI Inference Specialist SqueezeBits to Become Full-Stack AI Infrastructure Company

Korean AI hardware startup Rebellions acquired SqueezeBits, an AI inference specialist, to expand from hardware to full-stack infrastructure including software and serving capabilities. The move aims to create an end-to-end AI platform.

Schneider Electric agrees to acquire industrial AI firm Cognite

Schneider Electric announced an agreement to acquire Cognite, a data and analytics company specializing in industrial AI. The deal aims to enhance strategic intelligence for industrial operations, combining Schneider's domain expertise with Cognite's AI capabilities.

Brand Engagement Network Completes Acquisition of Cataneo

Brand Engagement Network (BEN) finalized its acquisition of Cataneo GmbH, a profitable software business. The acquisition expands BEN's enterprise AI platform with an established customer base and revenue stream.

Aikido acquires Root to patch open-source software without forced upgrades

Belgian cybersecurity firm Aikido Security acquired Root.io, an Israeli startup, for an estimated $70-$100 million. The deal integrates Root's technology for patching vulnerabilities without requiring upgrades, strengthening supply-chain security for AI systems.

AI product & feature launches

Meituan open sources LongCat-2.0, a 1.6 trillion parameter coding model trained on Chinese chips

Chinese delivery giant Meituan released LongCat-2.0, an open-source 1.6 trillion-parameter coding AI model trained entirely on domestic Chinese hardware. It demonstrates China's ability to train frontier AI without reliance on Nvidia GPUs, amid export controls.

Research with immediate practical relevance

Japan's shape-shifting drone achieves fully autonomous outdoor flight

Researchers at the University of Tokyo demonstrated a morphing aerial robot (DRAGON v1.5) able to fly autonomously using onboard sensors, changing shape mid-flight. The snake-like drone can navigate complex environments, advancing physical AI in robotics.