Key facts
- A UN report warns that the rapid advancement of artificial intelligence poses a risk of widening global inequality.
- Experts state the window for establishing global governance and shared rules for AI is closing quickly.
- The report highlights risks including AI-generated deepfakes, disinformation, and potential misuse for harmful purposes.
- A significant disparity exists in global AI computing capacity, with the US and China holding the vast majority.
- Developing nations lack the infrastructure and funding to develop or audit AI, exacerbating the inequality gap.
- The panel calls for enhanced international coordination, stronger third-party evaluation, and investment in AI governance.
A new report from the United Nations warns that the rapid advancement of artificial intelligence could significantly widen the gap between rich and poor countries, and that the window for establishing global governance to mitigate this risk is rapidly closing.
The UN's Independent International Scientific Panel on Artificial Intelligence, comprising 40 global experts, released a preliminary report highlighting the exceptional speed of AI development, particularly in generative and agentic AI. These systems are now capable of complex tasks, with their capabilities roughly doubling every few months, making them increasingly difficult to monitor and control without robust safeguards.
UN Secretary General António Guterres emphasized the urgency, stating, "The more AI advances without shared rules, the less say governments and people will have in the outcome." He urged governments not to delay action, as the scientific understanding of AI's potential is already established.
The report detailed significant risks associated with AI, including its use to generate child sexual abuse material and explicit deepfakes, disproportionately affecting women and children. It also noted AI's role in creating more convincing and harder-to-detect disinformation, eroding public trust and democratic discourse. Cybersecurity threats are escalating, with criminals leveraging AI for fraud and social engineering, while vulnerable users face risks to their mental health.
However, the report also acknowledged potential benefits, such as AI models mapping protein structures to accelerate drug discovery and vaccine research, and aiding in identifying food insecurity. AI is also seen as a tool for broadening access to education and mental health support.
A key concern raised is the uneven global distribution of AI capabilities. The United States reportedly controls about three-quarters of the computing power for leading AI supercomputers, with China holding around 15%, together accounting for 90%. This concentration means the most advanced AI models are developed by companies in these two nations, leaving developing countries without the necessary talent, infrastructure, and funding to build or audit the AI systems they use, thereby risking an increase in global inequality.
The panel also pointed to challenges in regulating AI, citing an "evidence dilemma" where AI evolves faster than data can be collected for effective rule-making. While over 40 AI governance frameworks exist globally, they are described as fragmented, inconsistent, and rarely tested. The reliance on the same companies developing the technology for safety testing raises independence concerns.
To address these issues, the panel advocates for stronger third-party evaluation, increased international coordination, shared standards, and investment to empower countries to govern AI on their own terms. The report's findings are set to feed into the UN Global Dialogue on AI Governance, scheduled to open in Geneva on July 6, 2026, where member states will discuss coordinated international approaches to managing AI.