OpenAI, Anthropic, and Block Form Agentic AI Foundation

OpenAI, Anthropic, and fintech company Block have teamed up to create the Agentic AI Foundation (AAIF), a group housed under the Linux Foundation that will develop standards for a growing class of AI systems known as agents. The initiative is backed by Google, Microsoft, Amazon Web Services, Bloomberg, and Cloudflare.

The move reflects growing concern across the industry that AI agents, which can carry out tasks autonomously, are being developed with varying frameworks and communication methods. In a blog post, OpenAI mentioned that as these systems take on more responsibility, fragmentation poses risks. “The cost of fragmentation increases,” the company wrote, warning that incompatible approaches could limit portability, safety, and broader progress.

Furthermore, the company stated that the Agentic AI Foundation is designed to serve as a neutral venue where rules for interoperability can be drafted, debated, and updated. The Linux Foundation, a nonprofit that oversees widely adopted open-source projects including the Linux operating system and Kubernetes, will provide governance and long-term management. 

According to OpenAI, this structure is designed to prevent any single company from controlling the direction of the standards.

Founding members are contributing their existing tools to the effort. OpenAI is submitting AGENTS.md, an open format for agent instructions. Anthropic is contributing its Model Context Protocol, used to connect AI applications. Block is adding Goose, its open-source agent framework.

The companies stated that the tools were already publicly available, but placing them under a shared foundation would enable outside developers to help refine and extend them.