Open-source Python toolkit for assessing and mitigating fairness issues in ML models. By 2026, Fairlearn is a common building block in responsible-AI pipelines,
Fairlearn is the open-source responsible-AI workhorse: fairness metrics across sensitive groups, mitigation algorithms, and assessment workflows that turn 'is the model biased?' into measurable engineering, a standard component of compliance-grade ML pipelines.
Pricing: Free and open-source (MIT).
It can measure disparities and apply mitigations against a chosen definition, but fairness definitions trade off against each other, so the human decision of which to satisfy remains the hard part it can't automate.
Directly to structured outputs/decisions; open-ended generation needs complementary techniques (prompt-based probes, output audits). For classic decision models, hiring, credit, it's the standard tool.