OpenAI announces the co-founding of the Agentic AI Foundation (AAIF) under the Linux Foundation, alongside Anthropic and Block, with support from Google, Microsoft, AWS, Bloomberg, and Cloudflare. This foundation aims to provide neutral governance for open, interoperable infrastructure as agentic AI systems move from experimentation to real production use.

The announcement comes at a pivotal moment: AI agents are evolving from prototypes into tools handling concrete responsibilities in enterprise and consumer settings. Without common conventions and neutral governance, agent development risks diverging into incompatible silos, limiting portability, security, and progress.

Each co-founder contributes a major project to the foundation. OpenAI brings AGENTS.md, a simple Markdown format for providing agents with project-specific instructions (coding conventions, build steps, test requirements). Launched in August 2025, AGENTS.md has been adopted by more than 60,000 open source projects and agent frameworks including Cursor, Devin, GitHub Copilot, Gemini CLI, and VS Code. Anthropic contributes the Model Context Protocol (MCP), an open protocol for integration between AI systems and external data sources. Block contributes goose, a local-first open source AI agent framework.

OpenAI highlights its prior contributions to the open agentic ecosystem: Agents SDK, Apps SDK, Agentic Commerce Protocol, gpt-oss models, and Codex CLI. The latter has enabled the merging of more than two million public pull requests on GitHub. The company has also integrated MCP as the foundation for connectors and apps in ChatGPT, demonstrating that open infrastructure can achieve real adoption.

The AAIF operates as a directed fund under the Linux Foundation, a nonprofit organization with decades of experience governing critical projects (Linux Kernel, Kubernetes, Node.js, PyTorch). This structure ensures that no single company controls the direction of foundational infrastructure alone.

AGENTS.md's contribution to the AAIF ensures that developers and companies have a portable way to provide context to agents, that the format can evolve openly with input from many tools and communities, and that no single company controls its direction as adoption grows.

This initiative marks a significant step in the maturation of the AI agent ecosystem, favoring industry collaboration over proprietary fragmentation.