OpenAI, Anthropic and Block form new body to standardise Agentic AI

OpenAI, Anthropic and Block have jointly launched the Agentic AI Foundation (AAIF), a new open-source organisation housed under the Linux Foundation to advance interoperable standards for agentic AI. The initiative has backing from Google, Microsoft, AWS, Bloomberg and Cloudflare, signalling strong industry alignment around shared infrastructure for the next wave of AI systems.

As part of the effort, the companies are contributing several widely used technologies to AAIF: Anthropic’s Model Context Protocol, which enables agents to connect and interact; OpenAI’s AGENTS.md format, which provides project-specific instructions for coding agents; and Block’s Goose framework for building agentic tools. While all three were already freely available, their transfer to a community-governed foundation opens the door for broader ecosystem participation and evolution.

AGENTS.md, launched in 2025, has already been adopted by more than 60,000 open-source projects and frameworks. Its move into AAIF is designed to ensure long-term, vendor-neutral stewardship as agentic systems transition from early experimentation to real-world deployment.

By consolidating key protocols and frameworks under one umbrella, AAIF aims to offer a neutral platform—supported by the Linux Foundation’s proven open-source governance—to maintain and advance interoperability standards, reduce fragmentation and enable scalable, safe adoption of agentic AI across industries.