Recently, Anthropic, Block, and OpenAI announced their partnership in an effort to further artificial intelligence interoperability. We’re excited that they are doing this initiative under the auspices of the Linux Foundation. This collaboration aims to standardize protocols for AI agents, ensuring that future developments in this field are open and accessible rather than dominated by proprietary technologies.
The announcement of this partnership is a watershed moment for the AI industry. Anthropic is donating its Model Context Protocol (MCP). This improved framework has made it much more efficient and simpler to connect these models, agents, tools, and data. To help you get started, Block is providing its open-source agent framework, Goose. In parallel, OpenAI is releasing AGENTS.md, a minimal instruction file optimized for AI coding assistants. Collectively, these contributions are poised to help establish a more collaborative and transparent AI ecosystem.
Linux Foundation Executive Director Jim Zemlin reaffirmed the importance of this initiative. That’s how we’ll avoid a future of “closed wall” proprietary stacks. The Linux Foundation Technical steering committees will map out technical roadmaps for the projects. This checking process prevents any individual member from having full autonomy over the course of the project. This kind of structure pushes everyone to make sure that they are contributing as equitably as possible.
“The main goal is to have enough adoption in the world that it’s the de facto standard,” stated David Soria Parra, reflecting the aspirations of the founding members. The AAIF (AI Agent Interoperability Foundation) is working to create the shared guardrails AI agents need. This initiative is a key step to make these systems trustworthy and scalable on a broad level.
Anthropic, Block, and OpenAI are accompanied by other notable members of the AAIF including AWS, Bloomberg, Cloudflare, and Google. Together, they seek to establish a standard framework that would enable AI agents to interoperate seamlessly across various platforms and applications.
Brad Axen from Anthropic emphasized the value of community participation in helping to refine these protocols, and we agree.
“Getting it out into the world gives us a place for other people to come help us make it better.” – Brad Axen
He added that the innovations brought by the open-source community would eventually benefit their own products. We agree with this sentiment, as it dovetails nicely with our larger intention to create a more participatory model for developing these protocols.
It’s a mindset,” OpenAI’s Nick Cooper said in agreement with this outlook. Most importantly, he emphasized the importance of developing many protocols to improve communication and collaboration between AI agents.
“We need multiple protocols to negotiate, communicate, and work together to deliver value for people,” – Nick Cooper
Cooper shared a hope that these protocols would be not a static element, but something that could evolve.
“I don’t want it to be a stagnant thing… They should evolve and continually accept further input.” – Nick Cooper
That commitment to evolution is the foundation of this program, which is essential given the fast pace of technological change we find ourselves in. This will include the AAIF’s interoperability and safety patterns. Therefore, this focus will require developing best practices for AI agents that enhance their reliability and effectiveness.
As Jim Zemlin, Executive Director of the Linux Foundation noted, ultimately success will be judged by the adoption of these standards. In addition, he stressed the necessity of these standards to enable collaboration between vendors across the globe.
“An early indicator of success, in addition to adoption of these standards, would be the development and implementation of shared standards being used by vendor agents around the world,” – Jim Zemlin
This joint commitment exemplifies the deep commitment that unites tech leaders. They understand that when government is wedded to private or proprietary solutions, it ultimately slows down government and AI innovation.