OpenAI, Anthropic and Block join new Linux Foundation effort to standardize the AI agent era

As artificial intelligence continues to evolve from simple chatbots to more sophisticated systems capable of taking real-world actions, the Linux Foundation is spearheading a new initiative designed to streamline the development of AI agents. This initiative, known as the Agentic AI Foundation (AAIF), seeks to prevent the fragmentation of AI technologies into incompatible, proprietary solutions. The AAIF is being launched with significant contributions from major industry players including Anthropic, Block, and OpenAI. Anthropic is providing its Model Context Protocol (MCP), which serves as a standard method for connecting models and agents to various tools and datasets. Meanwhile, Block is contributing its open-source agent framework, Goose, and OpenAI is sharing AGENTS.md, a straightforward instruction file that developers can incorporate into their repositories to guide AI coding tools. These contributions are considered foundational elements for a new era of AI agents. The AAIF also counts organizations such as AWS, Bloomberg, Cloudflare, and Google among its members, underscoring a collective industry effort to establish reliable standards that ensure trustworthiness in AI at scale. According to OpenAI engineer Nick Cooper, these protocols will act as a common language enabling different agents and systems to collaborate efficiently without necessitating that developers create integrations from scratch. “We need multiple protocols to negotiate, communicate, and work together to deliver value for people,” Cooper noted, suggesting that a singular provider is unlikely to dominate the landscape. Jim Zemlin, the executive director of the Linux Foundation, emphasized the importance of avoiding a future dominated by closed, proprietary systems. He stated, “By uniting these projects under the AAIF, we can coordinate efforts on interoperability, safety standards, and best practices specifically tailored for AI agents.” While Block is typically recognized for its fintech solutions, its introduction of Goose reflects a commitment to open-source alternatives. AI Tech Lead Brad Axen highlighted that Goose is already utilized by thousands of engineers weekly for tasks like coding and data analysis. “Making it open-source allows others to contribute to its improvement,” Axen explained, adding that this collaborative approach benefits Block as well. Anthropic’s decision to donate the MCP to the Linux Foundation is aimed at establishing it as a neutral infrastructure for connecting AI models to tools and applications, eliminating the need for custom adapters. MCP co-creator David Soria Parra expressed the hope that it will become the industry standard. The AAIF is structured to be funded through membership dues, but Zemlin reassured that financial contributions do not equate to control over project direction—this will be determined by technical steering committees. The success of the AAIF will be measured by the adoption of its standards and the actual implementation of shared protocols by AI agents globally. For Cooper, the evolution of these standards is crucial. “I don’t want it to be a stagnant thing. I want these protocols to adapt and grow,” he stated. There is also the concern that one company’s version could unintentionally become the standard simply by virtue of being first to market. Zemlin countered that this is a natural outcome of open-source development, where success is derived from quality rather than vendor dominance. For developers and enterprises, the immediate benefits are clear: reduced time spent on custom integrations, more consistent agent behavior across platforms, and easier deployment in secure environments. The broader vision is ambitious; if tools like MCP, AGENTS.md, and Goose gain traction, the AI agent ecosystem could transition from fragmented, closed systems to a cohesive, interoperable software landscape, reminiscent of the early web's modular architecture.