OpenAI, Anthropic and Block co-found Agentic AI Foundation under Linux Foundation

OpenAI, Anthropic and Block co-founded the Agentic AI Foundation, a new body under the Linux Foundation intended as neutral stewardship for open-source infrastructure used by AI agents. Each founder donated a project it had previously maintained independently: Anthropic contributed the Model Context Protocol (MCP), its standard for connecting models to external tools and data sources; OpenAI contributed AGENTS.md, a convention for giving coding agents project-specific instructions; and Block contributed Goose, an open-source, local-first agent framework. Amazon Web Services, Google, Microsoft, Bloomberg and Cloudflare joined as founding members alongside the three donors.

The stated purpose was to prevent the fast-growing agentic-AI ecosystem from splintering into incompatible, vendor-controlled stacks, as had partly happened with earlier generations of AI tooling. Linux Foundation executive director Jim Zemlin framed the goal as avoiding a “closed wall” of proprietary protocols; Block’s AI technical lead, Brad Axen, said Goose’s donation was meant to demonstrate that open alternatives could match proprietary agent products.

The Foundation’s significance rests on the identity of its founders rather than any single technical commitment: OpenAI and Anthropic are direct commercial rivals, and MCP in particular had already been adopted widely enough — including by Google and Microsoft — that placing it under neutral governance addressed a real concern about a de facto standard remaining under one company’s control. Whether the arrangement produces binding specifications or functions mainly as a coordination forum was not resolved at launch.