OpenAI, the company behind ChatGPT, announced on December 9 the creation of the Agentic AI Foundation (AAIF), an organization responsible for managing open source and interoperable infrastructure linked to artificial intelligence (AI) agents.
AAIF was founded by Anthropic and Block and supported by Google, Microsoft, Amazon, Bloomberg, and Cloudflare.
Additionally, the Agentic AI Foundation operates as a managed project within the nonprofit Linux Foundation. Manage open source projects, It serves as the core of the Linux operating system and specializes in neutral technology governance.
His role is to bring “proven expertise in neutral governance, community building, and long-term sustainability.”
The text states that “open standards make agents more secure, easier to build, and more portable across tools and platforms,” and as these technologies take on real-world tasks, “The cost of fragmentation increases.”.
Without common rules and neutral governance, the development of AI agents could lead to closed ecosystems competing with each other, which AAIF is trying to anticipate with a shared architecture from the beginning.
What is the purpose of Agentic AI Foundation (AAIF)?
According to the statement, AAIF aims to: Along with the development of autonomous agents By AI (a system that can perform tasks without direct supervision) From the experimental stage to the actual usage environment.
The aim is to evolve these technologies “in a transparent, collaborative and public interest-oriented manner” and to prevent the infrastructure supporting them from being under the control of a single company.
In practice, the AAIF serves as a central point for building interoperability standards, rules that allow different artificial intelligence systems to understand the same instructions and operate under a common standard.
The goal is to “establish a neutral, shared platform for building agent systems that benefit developers, enterprises, and the open community,” according to the announcement.
AGENTS.md: OpenAI’s contribution to a common platform
The statement details that OpenAI will contribute to the AGENTS.md standard, a file designed to guide agent behavior.
According to the text, it is designed to provide “a simple, interoperable source of project-specific guidance.” This includes: Programming conventions, software construction procedures, testing requirements.
In other words, this is a document that accompanies the traditional README.md, but is intended only for agents. Enabling AI systems Understand how to behave within a repository without relying on the specific rules of each tool.
AGENTS.md is part of the infrastructure. Although it is neither an AI model nor a programming tool, How to avoid unexpected agent behavior By AI Must operate in a complex environment.
OpenAI emphasizes that AGENTS.md has been adopted since AGENTS.md was published in August 2025. Over 60,000 open source projects Use tools like Amp, Codex, Cursor, Devin, Gemini CLI, Github Copilot, Jules, and VS Code.
The OpenAI team said the adoption reflects “increasing collaboration around shared, neutral practices.”
The Foundation will now be tasked with evolving this standard with the participation of multiple organizations. As OpenAI stated, their contributions ensure that “as adoption continues to grow, no company can control its direction.”
Can AAIF strengthen collaboration between AI and cryptocurrency networks?
The creation of AAIF opens up the possibility of connecting with Ethereum. More precisely, it uses the ERC-8004 standard, a recent proposal reported by CriptoNoticias and developed in collaboration with Google and Coinbase.
AAIF aims to establish common rules for the behavior of autonomous agents, and ERC-8004 proposes an identity and reputation chain layer to make this possible. See how these agents work and on what basis they carry out their duties.
The foundation can provide neutral governance and interoperability, and the standards Decentralized technical support for recording actionsvalidate your actions and avoid dependence on a single company.
If both initiatives proceed in a compatible manner, AAIF-defined agents could operate in a blockchain environment with verifiable identities and transparent accountability mechanisms, enabling an open market for AI-based services.
However, its relevance is potential, given the recent launch of the AAIF and the fact that Ethereum’s ERC-8004 standard is still in draft stage and needs to be discussed by the community.