Hermes Agent, developed by Nous Research, crossed 140,000 GitHub stars in under three months and is now the most used agent in the world according to OpenRouter. The framework is designed for reliability and self-improvement, provider-agnostic, and optimized for local use on NVIDIA hardware.

Source: NVIDIA Blog

Published May 14, 2026, this analysis shows that AI agent development has fundamentally shifted—SDKs now ship with built-in tools, the skills system replaced tool registries, and longer context windows reduced reliance on vector search, eliminating much of the middle infrastructure layer.

Source: Pulumi Blog

Microsoft achieved 1.0 release of Microsoft Agent Framework in April 2026, introducing a production-ready SDK with support for tool approval workflows, multi-agent orchestration, and human-in-the-loop controls for sensitive operations.

Source: .NET Blog

Multiple production-ready releases in the past week (May 8-13, 2026): Broadridge announced agentic capabilities for post-trade automation; SAP's Production Planning Agent becoming GA in Q2 2026; Notion launched Developer Platform with Workers and External Agent API; Dotmatics released Luma Agent; Sweet Security published agentic red-teaming product; Honeycomb introduced agent-native observability features.

Source: AI Agent Store

Analysis of the most-starred open-source AI agent frameworks as of April 2026, noting convergence around three open standards: Model Context Protocol (MCP) donated by Anthropic in December 2025, Agent2Agent (A2A) protocol from Google, and AGENTS.md convention from OpenAI.

Source: Pasquale Pillitteri

GPT-5.5 Instant became the default model for all ChatGPT users starting May 5th, 2026. OpenAI claims the model produces 52.5% fewer hallucinated claims on sensitive prompts compared to GPT-5.3 Instant, with a focus on more concise responses.

Source: mwm.ai

Microsoft's May 2026 update for Microsoft 365 Copilot integrates GPT-5.5 models alongside smarter Researcher capabilities, the return of the Waffle app launcher, and improved Copilot Notebooks features.

Source: The Windows Central

Coverage of 13 major AI stories from May 18, 2026 including Google I/O 2026, Anthropic's $900B funding round, GPT-5.5 Instant becoming the default, and other significant industry developments.

Source: Build Fast with AI

Google I/O 2026 on May 19-20 will showcase the next-generation Gemini 3.2 or 3.5 model with performance comparable to GPT-5.5, emphasizing on-device optimization capabilities.

Source: AIBase News

A critical NGINX vulnerability (CVE-2026-42945) disclosed last week is being exploited by attackers. The vulnerability, dubbed NGINX Rift, can be reliably exploited to trigger a denial-of-service condition and can potentially allow for unauthenticated remote code execution. VulnCheck's canary systems began flagging exploitation attempts on May 16.

Source: Help Net Security

May 2026 dropped three critical Linux vulnerabilities that together form a reliable, race-free, forensically quiet kill chain from the public internet to root. CVE-2026-42945 (NGINX Rift) is a heap buffer overflow in ngx_http_rewrite_module that gets remote code execution with no auth or prerequisites.

Source: Security Boulevard

Microsoft released security updates for 120 flaws with 17 "Critical" vulnerabilities, 14 of which are remote code execution, 2 are elevation of privilege, and 1 is an information disclosure flaw.

Source: BleepingComputer

CVE-2026-31431 ("Copy Fail") is a high-severity local privilege escalation vulnerability affecting the Linux kernel's cryptographic subsystem with a logic flaw within the algif_aead module. The exploit is 732 bytes of Python with no races, no disk writes, and works on every distro shipped since 2017.

Source: Microsoft Security Blog

CISA added CVE-2026-6973 (Ivanti Endpoint Manager Mobile improper input validation vulnerability) to the Known Exploited Vulnerabilities Catalog based on evidence of active exploitation.

Source: CISA

Released May 13, 2026, VS Code 1.120 brings the Agents window to Stable, improves BYOK model visibility and control, and adds Markdown quality-of-life improvements and agent safety features.

Source: code.visualstudio.com

VS Code shipped releases v1.116 through v1.119 throughout April and early May 2026, with Copilot now able to search by meaning in any workspace and run grep-style queries across GitHub repos and orgs. Bring-your-own-key support extends to Copilot Business and Enterprise.

Source: GitHub Blog

VS Code moved to weekly stable releases, with VS Code 1.116 featuring stronger chat and agent workflows, including agent debug logs, Copilot CLI thinking effort controls, terminal agent tools, and built-in GitHub Copilot.

Source: GitHub Blog

AWS launched Amazon Quick—an AI assistant for work with a desktop app and expanded integrations. Amazon Quick is an AI assistant for work that connects to all of them, learns what matters to you, and takes action on your behalf. AWS and OpenAI expanded partnership by bringing the latest OpenAI models including GPT-5.5 and GPT-5.4 to Amazon Bedrock in limited preview.

Source: Amazon Web Services

Powered by custom 6th-gen Intel Xeon Scalable processors, these instances deliver up to 43% higher performance over M6in and M6ib. M8in offers 600 Gbps network bandwidth, while M8ib delivers up to 300 Gbps EBS bandwidth.

Source: Amazon Web Services

Memory-optimized instances built on the same 6th-gen Intel Xeon Scalable processors and Nitro cards, with the same 600 Gbps network and 300 Gbps EBS bandwidth profiles. Well-suited for large commercial databases, data lakes, and in-memory databases such as SAP HANA.

Source: Amazon Web Services

Amazon Bedrock AgentCore previewed the first managed payment capabilities enabling AI agents to autonomously access and pay for APIs, MCP servers, web content, and other agents.

Source: Amazon Web Services

AWS announces the general availability of the AWS MCP Server, a managed remote Model Context Protocol (MCP) server that gives AI agents and coding assistants secure, authenticated access to all AWS services. The AWS MCP Server is part of the Agent Toolkit for AWS, a suite of tooling.

Source: Amazon Web Services

On May 14, 2026, Anthropic announced a $200 million partnership with the Gates Foundation committing grant funding, Claude usage credits, and technical support for programs in global health, life sciences, education, and economic mobility over the next four years.

Source: Anthropic

In mid-May 2026, Anthropic launched Claude for Small Business, integrating Claude into tools like QuickBooks, PayPal, HubSpot, Canva, Docusign, Google Workspace, and Microsoft 365 with ready-to-run workflows for payroll, invoicing, sales, marketing, and month-end close.

Source: Anthropic

Anthropic added Fast mode support for Claude Opus 4.7 in research preview, enabling faster output token generation. The feature can be accessed by setting speed to 'fast' with the claude-opus-4-7 model and the fast-mode-2026-02-01 beta header.

Source: Anthropic

On May 4, 2026, Anthropic announced a joint venture valued at $1.5 billion focused on deploying enterprise AI services, with Blackstone, Hellman & Friedman, and Goldman Sachs as founding partners alongside backing from multiple major investors.

Source: TechCrunch

On May 6, 2026, Anthropic announced a deal to use SpaceX's Colossus 1 facility in Memphis, Tennessee, providing 300 megawatts of new computing capacity with 220,000+ Nvidia processors and introducing a new 'dreaming' feature for Claude.

Source: Al Jazeera