Global Index on Responsible AI: 2026 Report

Policy expansion doesn't guarantee protection: While 126 of 135 countries (93%) have adopted at least one government AI policy or initiative, most frameworks lack enforcement mechanisms—particularly in the Global South, where 78% of new AI frameworks remain non-binding compared to 42% in the Global North.

Global South faces enforcement gap: Global South countries created 203 of 306 new AI governance frameworks since the first GIRAI report, but the majority lack legal teeth to actually protect citizens from harmful AI systems.

Governments don't practice what they preach: Despite Transparency and Explainability being identified as one of the strongest policy areas, governments often fail to apply these same standards to their own algorithms used in public services, creating a credibility gap between stated commitments and actual practice.

Responsible AI governance is still emerging: The assessment of 68,138 data points across 135 countries reveals that meaningful institutional capacity and redress mechanisms for AI harms remain underdeveloped globally, even as countries increasingly adopt responsible AI policies on paper.

Human rights frameworks matter but need implementation: The report grounds assessment in UNESCO's Ethics of AI Recommendation and other human rights standards, suggesting that translating these international commitments into enforceable national protections remains the critical challenge for responsible AI governance.