The Informed Layer · Newsscan
Daily briefing

Agentic workflows surge; Anthropic partnerships expand

← All issues

Anthropic / Claude ecosystem

Cognizant and Anthropic Expand Partnership to Embed Claude in Industry Platforms

Cognizant and Anthropic have announced an expanded partnership that sees Claude integrated across Cognizant's industry platforms. Cognizant has also been promoted to Global Premier Partner in Anthropic's Claude Partner Network, enabling a 'Frontier Certified' workforce model to accelerate enterprise AI deployments.

Frontier model providers

OpenAI Unveils GPT-5.6 Family with Three Variants and Enterprise Tools

OpenAI has announced GPT-5.6, a new family of models with three variants (Sol, Terra, Luna) that feature enhanced cybersecurity and coding capabilities. The release also includes ChatGPT Work, an enterprise-focused tool for deployment and management.

Cohere Launches North Automations for Agentic Workflow Orchestration

Cohere has introduced North Automations, a tool for orchestrating agentic AI workflows. The product enables enterprises to design, test, and deploy complex AI-driven process automations using Cohere's models.

AI developer tooling & infrastructure

Arcade.dev Ships MCP with Stateless Core and Hardened Auth

Arcade.dev has released a production-ready version of the Model Context Protocol (MCP) with a stateless core, hardened authentication, and first-class support for extensions. The release is designed to simplify scaling for AI agent integrations.

You.com MCP Server Comes to Warp Terminal

You.com has integrated its MCP server into the Warp terminal, allowing AI assistants to access live web search and contextual data directly within the developer environment.

Cloud & platform providers

No significant new developments.

AI policy, regulation & governance

EU AI Omnibus Enters into Force with Extended Timelines

The EU's AI Omnibus regulation has officially entered into force, introducing extended implementation timelines while preserving key safeguards introduced by the original AI Act. The law harmonises and clarifies requirements for AI systems across the bloc.

Safe Work Australia Publishes AI and Digital Technology Safety Guidance

Safe Work Australia has issued new guidance on managing workplace health and safety risks from AI and digital technologies, integrating these tools into standard WHS risk management processes with an emphasis on consultation and existing duty-holder frameworks.

Industry & market moves

Ode with Anthropic and LogicGate Partner for Agentic GRC Capabilities

Ode, a joint venture of Anthropic and Blackstone, has partnered with LogicGate to accelerate agentic AI capabilities for governance, risk, and compliance. The collaboration aims to scale expert-led client outcomes by integrating Claude into LogicGate's risk platform.

Samsung SDS Partners with Anthropic for Claude Adoption in Korea

Samsung SDS has formed a strategic partnership with Anthropic to expand Claude adoption among Korean enterprises and drive the AI-native transition for Samsung affiliates. The deal aims to combine Samsung’s enterprise reach with Anthropic’s frontier models.

Ilya Sutskever’s Safe Superintelligence Partners with Nvidia to 10x Compute

Safe Superintelligence Inc. (SSI), founded by Ilya Sutskever, has entered a strategic partnership with Nvidia that includes a multi-billion dollar investment and access to Nvidia's Vera Rubin computing platform. The deal aims to scale SSI's compute capacity by an order of magnitude for safe superintelligence research.

Enigma Raises $70M Seed to Develop Foundation Models for Intuitive Robot Control

Israeli robotics startup Enigma emerged from stealth with a $70M seed round to build AI foundation models that make controlling robots as simple as adjusting a volume knob. The company aims to study human-robot interaction and create a new kind of robotic brain.

AI product & feature launches

Claude Code Powers New Open-Source AI Job Application Assistant Framework

An open-source 'ai-job-search' framework has been released that uses Claude Code to manage structured, verifiable AI-assisted job applications. It features a second reviewer agent to ensure quality and consistency.

Wiz Introduces Atlas, an AI Vulnerability Researcher for Cloud Security

Cloud security company Wiz has launched Atlas, an AI-powered vulnerability researcher that uses a multi-model, adversarial workflow to find and validate security weaknesses. It employs deterministic orchestration and a purpose-built harness for accuracy.

IAG Pilots Agentic AI for Customer Claims Using OpenAI Presence

Insurance Australia Group (IAG) is trialling OpenAI's Presence platform to handle customer-facing claims processing, a high-conduct-risk area. The pilot involves AI agents making decisions within a heavily governed framework.

Celeris Unveils Diffusion-Based LLM Enabling Real-Time AI with Near-GPT-5 Performance

Celeris has released Celeris-1, a diffusion-based large language model that promises ultra-low latency and performance approaching GPT-5. The technology could enable new real-time AI applications that were previously infeasible with autoregressive models.

Microsoft Launches First Cybersecurity Model and Agentic Defense Platform

Microsoft has released MAI-Cyber-1-Flash, its first dedicated cybersecurity model, alongside Project Perception, an agentic defense platform. The system uses multiple models to cut enterprise security costs by about 50% while achieving strong vulnerability detection.

Nvidia Launches Open Secure AI Alliance with Palantir, IBM, and Hugging Face

Nvidia has founded the Open Secure AI Alliance, bringing together Palantir, IBM, CrowdStrike, SpaceX, Hugging Face and others to develop and share open-source AI security tools. The initiative follows a high-profile OpenAI cyberattack and aims to harden open-weight models against threats.

Research with immediate practical relevance

Study Finds AI Chatbots Citing Russian Propaganda from EU-Sanctioned Outlet

A new study reveals that several large language models cite material from an EU-sanctioned Russian propaganda outlet and the r-FBI narrative, raising concerns about LLMs propagating disinformation even when trained on seemingly diverse data.

Moonshot AI Releases Kimi K3 Weights, the World's Largest Open-Source Model

Moonshot AI has released the weights for its 2.8 trillion parameter Kimi K3 model under a revenue-tiered license, fulfilling a previously announced plan. The model matches or exceeds many proprietary systems, sharply narrowing the open-source frontier gap.