Menu

Artificial intelligence (AI) is spreading through public services and daily life far faster than governments are building the institutions needed to safeguard human rights, according to a new global assessment released in Cape Town on 8 July.

The second edition of the Global Index on Responsible AI (GIRAI), produced by the South Africa‑based Global Centre on AI Governance, compares 135 countries and finds that while national AI strategies, laws and global commitments are increasing, enforcement capacity, oversight systems and transparency rules are not keeping pace.

The report says many governments are adopting responsible AI principles but lack the institutions, disclosure requirements and redress mechanisms needed to protect people as AI becomes embedded in welfare systems, healthcare, policing, education, finance and workplaces.

Rachel Adams, founder and CEO of the Global Centre on AI Governance, said the findings show a persistent mismatch between political commitments and real‑world capability. “Responsible AI cannot be secured through principles alone… Governments need enforceable obligations, independent oversight, public disclosure, monitoring systems and accessible routes for redress.”

Key findings of the study highlight several gaps. Showing that AI is outpacing governance, 53% of the global population has used generative AI tools, yet the average GIRAI score remains low at 35 out of 100. Evidence of implementation of AI governance measures exists in only 55% of cases, dropping to 45% in Global South countries.

Responsible AI content in Global South frameworks rose to 88% since the first edition, but 78% of these frameworks are non‑binding. Governments are prioritising technical safeguards while human harms remain under‑addressed, and misuse of AI by governments was found in 35 countries.

The findings also show that while transparency is rising, governments are not disclosing their own AI use, with only 18% of countries requiring disclosure of government algorithmic systems. Gender issues are increasingly recognised, but protections remain weak: 29 new countries added gender‑related AI frameworks, but fewer than half show evidence of implementation.

On the future impact of AI, the study finds that children are being prepared for the AI economy but not protected from AI‑related harms. AI literacy is widespread, yet only 41% of countries have frameworks addressing children’s rights in AI. Environmental impact remains a blind spot, with only 27% of countries addressing AI’s environmental footprint, and most frameworks are non‑binding.

The report also finds that local‑language AI is recognised but not required; AI skills are rising, but workers’ rights are neglected; and global governance is fragmenting. GIRAI scores average 55 in Global North countries and 27 in Global South countries, and only 36 countries have mechanisms for civil society participation.

GIRAI evaluates countries across three pillars – AI Policy, Civil Society Engagement, and Enabling Conditions – and five human‑rights‑linked dimensions: AI Use in Public Service Delivery, Ethics and Sustainability, Inclusion and Diversity, Labour and Skills, and Trust and Safety.

The index is built on global commitments, including the UNESCO Recommendation on the Ethics of AI and the OECD AI Principles, translated into measurable indicators.

Out of 39 African countries surveyed, only six score above the global average: Nigeria, Egypt, Kenya, Ghana, Benin and Morocco. North Africa is the continent’s strongest‑performing subregion.

The report concludes that responsible AI governance is expanding worldwide but remains fragmented, under‑enforced and insufficiently grounded in public accountability. Without a shared rights‑based floor, it warns, global interoperability risks serving markets before it protects people.