Introducing the FAIR² Open Specification

FAIR
AI-Ready
Responsible AI & Verifiable
Context-Rich
FAIR² (pronounced FAIR squared) formalizes the FAIR data principles into a verifiable, machine-actionable framework for AI-ready, responsibly governed data. It provides a context-rich structure linking methods, provenance, contributors, and governance in a form readable by both humans and machines—ensuring rigor, transparency, and reproducibility.

Built on open web standards and compatible with MLCommons Croissant, FAIR² integrates seamlessly with TensorFlow, JAX, PyTorch, Kaggle, and Hugging Face, enabling trusted, reusable, and interoperable data across disciplines.

Now online! The first public release of the FAIR² Open Specification — with thanks to all pilot participants for their contributions and insights.
Learn more About FAIR²

Join the Alliance

Please complete a brief survey to join the FAIR2 community.

What is FAIR²?

FAIR² (pronounced FAIR squared) formalizes the FAIR data principles into a verifiable, machine-actionable framework for AI-ready, responsibly governed, and context-rich data. It ensures that every dataset carries the structure, provenance, and clarity needed for transparent, ethical, and reproducible reuse across disciplines.
Context-Rich Metadata: Describe not just the data, but how, why, and by whom it was created and validated.
AI-Ready Design: Enable seamless integration into modern machine learning and data-driven workflows.
Responsible & Verifiable: Encode governance, consent, and provenance in computable form to ensure ethical, auditable reuse.

Why Choose FAIR²?

FAIR² advances the FAIR principles for the age of AI, making data not only findable and accessible but also verifiable, governed, and ready for machine interpretation.
Discoverability with Depth: FAIR² datasets are easy to find and easy to trust—each carries rich, structured context describing origin, purpose, and use conditions.
Trust Through Transparency: Detailed provenance, methodology, and ethical context make every dataset auditable and bias-aware, enabling confident reuse in both science and AI.
Built for Interoperability: Grounded in open web standards and linked-data principles, FAIR² connects existing data models into a unified, AI-compatible ecosystem for transparent, cross-domain collaboration.

Development Roadmap

FAIR² is developed in collaboration with the scientific and data community, evolving through shared goals of transparency, reproducibility, ethical alignment, and AI readiness.
Public Release: The first FAIR² Open Specification is now available (October 13, 2025).
Next Steps: Establish community governance and certification processes, while engaging new collaborators to extend FAIR² with additional capabilities and domain-specific requirements.
Mailing Address
FAIR² Alliance

Avenue Tribunal-Federal 34
1005 Lausanne,
Switzerland
Terms & Conditions
Pages
About
Specification
Get Involved
Contact
Contact
info@fair2.ai