The report’s central claim is that the ground is moving under the whole AI conversation: the shift from generative AI to agentic AI, systems that act, decide and coordinate with limited human intervention. It maps that shift across sectors tied to the Sustainable Development Goals, and against the governance scaffolding now being built to contain it.
The headline
Alongside agentic AI, the report foregrounds two ideas gaining traction with governments: sovereign AI, where states invest in domestic capability to reduce dependence on foreign providers, and its link to digital public infrastructure, the digital identity, payment and data-exchange systems that can make AI services cheaper and more inclusive to deliver. It surveys AI applied to education, health, climate and environment, infrastructure and agriculture, the applied core of the AI for Good agenda, and pairs the opportunity with the risks: bias, misinformation, loss of human control in autonomous systems, and the rising energy demand of data centres.
Why it matters
The report doubles as a map of the AI governance moment. Its governance chapter covers the EU AI Act and, closer to this coverage, the UN bodies stood up in 2025: the Global Dialogue on AI Governance and the Independent International Scientific Panel on AI, both flowing from the Global Digital Compact. Its SDG framing is organic rather than stamped on: AI is examined where it touches health (SDG 3), education (SDG 4), energy and climate (SDG 7, SDG 13) and food (SDG 2), with the honest caveat that the benefits are not automatic.
Watch & read
- AI for Good Impact Report, 2nd edition, the full report (ITU).
- AI for Good Global Summit 2026, the platform behind the report.
- Unlocking AI's Potential to Serve Humanity, the companion theme report.
The report labels itself a 2nd edition with 2025 content and carries a 2026 ITU copyright. Quotation is from the report’s foreword.