Google hands its A2A agent protocol to a neutral foundation
The rules for how AI agents talk to each other now sit alongside Anthropic's own connection standard under one independent roof.
The answer
Google moved its Agent2Agent protocol to the Linux Foundation's Agentic AI Foundation on 27 August.
If you've used one company's AI assistant and wished it could hand off a task to a different company's tool without breaking, this news is about the plumbing that makes that possible.
What happened
Google has transferred its Agent2Agent (A2A) protocol, a set of open rules that let AI agents built by different companies find each other, communicate and pass tasks along, to the Agentic AI Foundation (AAIF). AAIF is a Linux Foundation body set up to run open infrastructure for AI agents.
The move was announced on 20 August, and A2A's own blog confirmed on 27 August that the protocol had been accepted as a "Growth Stage" project, a formal status within the foundation. A2A launched in 2025 and is already backed by more than 150 organisations.
AAIF already hosts Anthropic's Model Context Protocol (MCP), along with two other projects, goose and AGENTS.md. The foundation's membership has grown fast: fewer than 40 companies at its December 2025 launch, more than 250 now, including AWS, Anthropic, Block, Bloomberg, Cloudflare, Google, Microsoft and OpenAI.
The two main protocols do different jobs. MCP connects an individual agent to outside tools and data, what the industry calls a "vertical" link. A2A lets separate agents work with each other as peers, a "horizontal" link. Together they cover most of how an AI agent reaches out beyond itself.
Google has compared its approach to Android: publish open standards, then let an ecosystem build on top of them to speed up adoption. Mazin Gilbert, AAIF's executive director, said no single company will build every agent an enterprise needs, according to the A2A blog.
Governance sits with a Governing Board, which handles strategy, budget and membership, and a Technical Committee, which reviews projects. The Linux Foundation itself only provides legal and operational structure, not technical direction, according to Techzine.
What it means for you
If you rely on AI agents through work or personal tools, this reduces the risk of getting locked into one company's ecosystem. Shared, neutral standards mean developers spend less time writing custom code to connect agents from different vendors, and more systems that combine tools from several companies become practical rather than a one-off engineering project.
What happens next
The notes don't say when new features or products built on the combined standards will appear. What's confirmed is that A2A now sits alongside MCP under the same independent foundation, with a governance structure designed to keep future development open rather than controlled by one company.
Sources
- A New Chapter for A2A: Joining the Agentic AI Foundation — A2A Protocol, 27 August 2026
- Google transfers A2A to the Agentic AI Foundation — Techzine, 17 August 2026