Fairmind

FairMind is an open-source platform for AI governance and assurance, currently in development. It aims to help organisations evaluate AI systems for bias and safety, and to collect reliable evidence for governance frameworks. The platform began as a tool for testing machine learning models for bias. It is now being redesigned to produce reproducible, reviewable evaluation evidence for a wider range of systems, including LLMs, AI agents and multimodal systems. The current version is an internal alpha, and the developers state that it does not provide compliance certification or automatic approval.

The platform consists of a web application and an application programming interface (API). Organisations register their AI models and define what is being evaluated and how. Existing features, retained from earlier versions but not yet independently validated, test predictive models against fairness metrics such as demographic parity, equalised odds and disparate impact. Other legacy features generate example code for reducing bias, such as reweighting data or adjusting decision thresholds. Results can be logged to tools such as MLflow and Weights & Biases.

The new assurance foundation records each piece of evaluation evidence with its exact scope, source and review status. Evidence can be signed and checked for authenticity, and it expires when it is no longer current. Accepted findings can be mapped to governance frameworks, with planned mappings to the EU AI Act, ISO/IEC 42001, the NIST AI RMF and India's DPDP Act. The backend is written in Python and the frontend in TypeScript. The code is released under the MIT licence.

Auto-discovered on 2026-02-11 by OECD Catalogue Automation