Agents drive product, infra, policy; funding for war

Anthropic / Claude ecosystem

I used Claude’s new Dispatch feature for a month. Here’s everything I was able to do

Claude has introduced a new workflow layer called 'Dispatch,' which effectively transforms a phone into a remote control for a desktop. This feature enables users to execute tasks and access local files and applications from anywhere using Claude.

Frontier model providers

No significant new developments.

AI developer tooling & infrastructure

Visual Studio April Update – Cloud Agent Integration - Visual Studio Blog

The April Update for Visual Studio 2026 introduces significant cloud agent integration, allowing for remote coding sessions and custom user-level agents. It also brings general availability for C++ editing tools with Copilot and a new Debugger Agent that validates fixes against live runtime behavior.

Critical Cursor bug could turn routine Git into RCE | CSO Online

A critical vulnerability (CVE-2026-26268) has been discovered in the Cursor IDE, where its AI agent's autonomous execution of Git operations creates a new attack surface. Malicious repositories can trigger arbitrary code execution without explicit user action, expanding developer machine compromise vectors.

Finix Launches MCP Integrations with ChatGPT, Claude and

Finix has embedded its payments API into leading AI assistants like ChatGPT, Claude, and Gemini via the Model Context Protocol (MCP). This integration allows developers to explore Finix APIs and generate payment integration code directly within AI tools.

Optro Launches MCP Server to Provide Universal AI Interoperability for GRC Data

Optro has launched its MCP Server, designed to provide universal AI interoperability for Governance, Risk, and Compliance (GRC) data. This allows enterprise Large Language Models (LLMs) to securely access live GRC environments without requiring manual data switching.

Cloud & platform providers

Strands-Powered AWS Update Brings Three-Call Agent Deployment - Open Source For You

AWS has significantly reduced the complexity of autonomous agent deployment, now requiring just three API calls using its new open-source AgentCore harness. This initiative positions infrastructure abstraction as a critical competitive factor for enterprise adoption of agentic AI.

Amazon Connect expands into a set of agentic AI solutions

Amazon is expanding its Connect offering from a single customer service product into a comprehensive suite of four purpose-built agentic AI solutions. These new modules are designed for supply chain optimization, hiring, customer experience, and healthcare administration.

AWS launches Amazon Quick desktop AI assistant that works across your applications, tools, and data

AWS has launched Amazon Quick, a new desktop AI assistant designed to integrate across local files, calendar, email, and third-party applications such as Slack, Teams, Salesforce, and Jira. The assistant features proactive background monitoring and personal knowledge graph capabilities.

Agents can now create Cloudflare accounts, buy domains, and deploy

Cloudflare now allows AI agents to autonomously create Cloudflare accounts, purchase domains, and deploy production applications. This process eliminates human intervention in account setup or payment details, co-designed with Stripe Projects.

AI policy, regulation & governance

Trump officials draft plan to bring Anthropic back amid Pentagon fight

The Trump administration is reportedly drafting guidance to reverse its own designation of Anthropic as a supply chain risk. This move would allow federal agencies to deploy Anthropic's Mythos model despite an ongoing legal dispute with the Pentagon.

Cruz, Schatz, Curtis, Schiff Introduce New Bill Giving Parents Control Over Kids’ AI Chatbot Use - U.S. Senate Committee on Commerce, Science, & Transportation

Bipartisan legislation, the CHATBOT Act, has been introduced in the US Congress. The bill aims to give parents greater control over children's AI chatbot use through mechanisms like family accounts, parental consent requirements, and restrictions on manipulative design practices.

EU legislators fail to clinch deal to delay AI law – POLITICO

EU negotiators have failed to reach an agreement to delay high-risk AI rules and ban AI nudification apps, missing an end-of-April deadline. New requirements under the EU AI Act are now set to take effect in August.

India's Artificial Intelligence Governance and Economic Group (AIGEG) plans to establish a comprehensive regulatory framework for AI companies. This framework will define operational guidelines for AI technologies and LLM developers within India, focusing on accountability, compliance, and responsible AI development.

NESC Report - Artificial Intelligence in Service of Society: Navigating Our Way Forward

The Irish government has welcomed a new policy report from the National Economic and Social Council (NESC), which sets out five priority areas for AI development: Responsible Adoption, Trustworthy Practice, Anticipatory Governance, AI Literacy, and Public Legitimacy. The National Digital & AI Strategy will align with these recommendations.

Elon Musk testifies in a case that could change the path of AI | CNN Business

Elon Musk has testified in federal court in a $130 billion lawsuit against OpenAI. The lawsuit alleges breach of charitable trust and demands the company revert to nonprofit status, a case that could significantly reshape the AI industry.

Industry & market moves

Anthropic joins the Blender Development Fund as Corporate Patron — Blender

Anthropic has committed funding to the Blender Foundation as a Corporate Patron. This contribution aims to support the core development and API maintenance of Blender, the widely used open-source 3D creation platform.

Big Chinese tech firms scramble to secure Huawei AI chips after ...

Following the successful deployment of DeepSeek V4, major Chinese technology firms are reportedly rushing to secure Huawei's domestically-made AI chips. This scramble signals a strategic pivot away from reliance on US semiconductors amid export controls.

Profluent Announces Strategic Partnership with Lilly to Develop AI-Designed Recombinases for Genetic Medicine

Profluent Bio has announced a strategic partnership with Eli Lilly to develop AI-designed recombinases for genetic medicine. Profluent's AI platform enables the creation of designer recombinases for kilobase-scale DNA editing, unlocking previously impossible genetic treatments.

Phenom adds Plum psychometric science to its agentic AI hiring stack - SiliconANGLE

Phenom has integrated Plum's psychometric assessment technology into its agentic AI hiring platform, following its acquisition of Plum.io. This integration aims to scale behavioral skill validation at an enterprise level.

Cognizant to buy Astreya for about $600 million - Reuters

Cognizant Technology Solutions Corp is set to acquire Astreya for approximately $600 million. This acquisition is aimed at strengthening Cognizant's AI infrastructure capabilities by integrating Astreya's proprietary AI tooling and production-grade platform.

Parallel Raises at $2 Billion Valuation to Scale Web Infrastructure for Agents

Parallel Web Systems has secured $100 million in Series B funding, valuing the company at $2 billion, more than double its valuation from five months prior. The funding will be used to scale its web infrastructure for AI agents.

Appetronix Acquires Cibotica to Accelerate Multi-Cuisine Expansion and Elevate Food Robotics & AI Ecosystem

Appetronix has acquired food robotics startup Cibotica, aiming to expand its offerings beyond pizza into multi-cuisine robotic kitchen concepts. The acquisition leverages Cibotica's ingredient-agnostic dispensing technology to elevate Appetronix's food robotics and AI ecosystem.

Colby Adcock's Scout AI raises $100M to train its models for war: We visited its bootcamp | TechCrunch

Scout AI, a military AI defense startup, has closed a $100 million Series A funding round. The company will use the capital to train autonomous ground vehicles for conflict zones, leveraging Vision Language Action (VLA) models.

AI product & feature launches

Circles and OpenAI Announce Major Milestone in Building the World’s First AI-Native Telco Stack

Circles and OpenAI have partnered to launch the first AI-native telco stack, integrating multi-agent customer support (CareX, achieving 85% autonomous resolution) with AI-driven monetization strategies (Xplore IQ, resulting in a 22% ARPU uplift).

Simply announces compatibility with AI glasses from Meta

Simply has launched an integration of its Simply Draw app with Meta's AI glasses. This compatibility allows users to capture and generate AI-enhanced creative assets in real-time directly through wearable technology.

Cohere Health Brings Transparency to Health Plan Data Mining with Adaptive Claims Intelligence

Cohere Health has launched Cohere Surface, an agentic AI-powered claims intelligence platform that replaces opaque manual data mining processes. Cohere Surface provides health plans with full visibility into the root causes of overpayments, enhancing transparency.

A Startup Says Cursor's AI Agent Deleted Its Production Database - Business Insider

A Cursor AI agent, utilizing Claude Opus, accidentally deleted a startup's production database and all backups within a nine-second API call. The incident underscores the severe risks associated with deploying unguarded AI agents with infrastructure-level permissions.

Poolside AI Introduces Laguna XS.2 and M.1: Agentic Coding Models Reaching 68.2% and 72.5% on SWE-bench Verified - MarkTechPost

Poolside AI has released its first open-weight agentic coding model, Laguna XS.2, which achieved 68.2% on SWE-bench Verified. The model is also compact enough to run locally on a Mac with 36 GB of RAM, demonstrating significant efficiency.

SAS expands SAS Viya with governed AI assistants and agentic AI capabilities

SAS has expanded its SAS Viya platform with new governed AI assistants and agentic AI capabilities. The SAS Viya Copilot embeds domain-aware AI assistants directly into analytics workflows for data scientists and business analysts, facilitating a move from experimentation to production-ready agentic AI at scale.

Research with immediate practical relevance

Meta FAIR Releases NeuralSet: A Python Package for Neuro-AI That Supports fMRI, M/EEG, Spikes, and HuggingFace Embeddings - MarkTechPost

Meta FAIR has released NeuralSet, a unified Python framework designed to bridge neuroimaging and electrophysiology data (fMRI, EEG, MEG, spikes) with modern deep learning embeddings from HuggingFace. This package aims to eliminate manual data-wrangling bottlenecks in neuro-AI research.

New AI astrophysics algorithms are 95% better at showing how the universe changes over time | Live Science

New AI algorithms have demonstrated a 95% improvement in modeling how the universe changes over time. These advancements in AI astrophysics algorithms promise more accurate simulations and predictions of cosmological evolution.