Annual Governance Report Harps on Steering the Future of AI

Emma Okonji

The International Telecommunications Union (ITU), the United Nations specialised agency for global digital technology development, has released the Annual AI Governance Report for 2025, which captures key insights ahead of next year’s Global Dialogue on AI Governance and consider how technical standards can shape Artificial Intelligence (AI) for positive global impact.

The report highlights the outcomes of the International AI Standards Exchange, which took place during the AI for Good Global Summit 2025 organised by ITU in Geneva, Switzerland in July this year.

The report, which measured what matters and how to assess AI’s environmental impact, also offered a comprehensive overview of current approaches to evaluating the environmental impacts of AI systems. The review focuses on identifying which components of AI’s environmental impacts are being measured, evaluating the transparency and methodology soundness of these measuring practices, and determining their relevance and actionability.

Highlighting the findings from academic studies, corporate sustainability initiatives, and emerging environmental tracking technologies, the report examines measurement methodologies, identifies current limitations, and offers recommendations for key stakeholder groups, such as developers (producers), users (consumers), and policy-makers.
One of the most pressing issues uncovered is the widespread reliance on indirect estimates when assessing energy consumption during the training phase of AI models. These estimates, according to the report, often lack real-time, empirical measurement.

The report further said important lifecycle stages remained significantly underexplored, adding that reliance on proxies introduces substantial data gaps, impedes accountability, and restricts consumers’ ability to make informed, sustainable choices about AI.

Giving details about the report, ITU Secretary-General, Doreen Bogdan-Martin, said: “As the global AI race continues to accelerate, humanity stands at a unique, transformative moment. Our collective challenge is not whether to govern artificial intelligence, but to understand and ensure governance steers AI in the right direction. This is at the heart of ITU’s mission to offer a neutral, global platform for artificial intelligence where everyone has a voice and sit at the table. Our second annual AI Governance Dialogue provided a timely opportunity for exactly this kind of multi-stakeholder discussion among governments, the private sector, academia, civil society organiaations, the technical community, and United Nations colleagues – each of whom has a key role to play.”

According to her, the ‘Ten Pillars for AI Governance’ captures the co-chairs’ vision for AI while reflecting the commitment of diverse stakeholders to translate dialogue into action for an AI that can deliver benefits for everyone, everywhere.

“We gladly offer this report, and in particular the conclusions of the co-chairs of this year’s AI Governance Dialogue, as input to Global Dialogue on AI Governance next year. Dialogue helps us set direction and chart a path towards a better inclusive digital future. We do not need to sail in the same ship, or at the same speed, but we do need to navigate the same oceans by the same compass,” Bogdan-Martin further said. 

The ITU’s Annual AI Governance Report 2025 emphasises the importance of multi-stakeholder dialogue in shaping AI governance frameworks.

The report highlights the need for inclusive governance to steer AI development positively.

With the emergence of AI agents, the report noted that the rise of AI agents has introduced new governance challenges and opportunities for productivity.

“AI agents can autonomously perform multi-step tasks with minimal human input, marking a shift from traditional AI tools. Current capabilities show mixed performance, with agents succeeding in simple tasks but struggling with complex ones. Policymakers face a narrow window to establish governance structures that address safety, fairness, and economic impact,” the report stated.

Addressing the issue of infrastructure, the report highlighted that the infrastructure for AI would continue to evolve rapidly, with significant implications for energy consumption. The report highlighted the uneven distribution of AI infrastructure globally, with significant gaps in regions like Africa and South America.