Standards are the bridge between AI principles and practical implementation. They operationalize abstract values into testable metrics, certification schemes, and technical specifications.

Advancing Global AI Governance Through Standards – Key Takeaways from the AI for Good Global Summit 2025

by ChatGPT-4o

Introduction

The AI for Good Global Summit 2025, hosted by the International Telecommunication Union (ITU) in collaboration with ISO, IEC, and over 50 UN agencies, brought together more than 11,000 delegates from 170 countries to explore how AI standards can serve as the bedrock of a trustworthy, inclusive, and sustainable AI ecosystem. With a dedicated “International AI Standards Exchange” track and 13 thematic workshops, the summit offered an expansive view of the current state of AI governance through standardization. This essay outlines the most surprising, controversial, and valuable findings from the summit and concludes with concrete recommendations for stakeholders.

Most Surprising Findings

1. Brain-Controlled Interfaces Now Reality
   Rodrigo Hübner Mendes drove a Formula One car using only his brainwaves. This was not a simulation but a real-life demonstration using non-invasive BCI technology. It’s symbolic of how fast frontier technologies are moving from science fiction to mainstream assistive applications.

2. AI Energy Footprint on Par with National Emissions
   AI’s contribution to greenhouse gas (GHG) emissions is now significant enough to require measurement standards akin to those used in national energy reporting. This is particularly pressing given the explosive rise of generative AI and hyperscale data centers.

3. AI-Specific Climate Impact Reports Announced by ITU and UNESCO
   Several institutions released dedicated frameworks to assess AI’s environmental impact, including ITU’s “Measuring What Matters” and UNESCO’s “Smarter, Smaller, Stronger”, signaling a new era of AI environmental accountability.

4. AI Governance Modeled After Internet Infrastructure
   A keynote highlighted that AI governance should mirror the distributed, resilient governance models used in internet systems, such as ICANN and IETF, suggesting a pivot from centralized regulatory paradigms.

5. AI Standards Database Now Publicly Available
   A searchable database covering over 700 AI-related standards was launched. It categorizes standards by industry, use case, and technical status, allowing real-time transparency and reducing duplicative efforts.

Most Controversial Statements

1. Multistakeholderism vs. Sovereignty
   Participants agreed on the need for interoperability and shared standards but warned of the risk to national regulatory sovereignty—especially as AI agents begin communicating autonomously across borders.

2. Standards as Regulatory Shortcuts?
   Some panelists implied that standards might effectively bypass slow legislative processes, enabling industry-led governance. This raised concerns about accountability, especially in high-risk sectors like health, mobility, and education.

3. Lack of Inclusion in Standards Development
   Despite calls for inclusivity, current standardization efforts still suffer from underrepresentation of the Global South, Indigenous voices, and those without technical expertise, creating a risk of cultural and ethical blind spots.

4. Use of Open-Source AI with Bias Metrics
   The idea of pairing open-source AI models with “bias and sustainability metrics” as a regulatory workaround generated debate about the feasibility and risks of pushing responsibility onto developers and end-users instead of platforms.

Most Valuable Insights

1. Standards as Translators of Ethics into Code
   The consensus was clear: standards are the bridge between AI principles (like transparency and fairness) and practical implementation. They operationalize abstract values into testable metrics, certification schemes, and technical specifications.

2. Synergy Between Regulation and Standards
   Participants urged co-development of regulation and standards. Standards should not be seen as optional or separate from legal frameworks, but as integrated tools that aid compliance while ensuring agility in tech deployment.

3. Emerging Domains: Multimedia Authenticity & Deepfakes
   Two key documents from the AI and Multimedia Authenticity Standards Collaboration were unveiled to tackle misinformation and deepfakes, including a regulatory checklist and a technical mapping of over 35 standards.

4. Sustainable AI at the Core of Future Frameworks
   Discussions on AI’s energy use emphasized the role of AI in achieving net-zero goals—both as a contributor to emissions and a potential accelerator of energy optimization through intelligent systems and edge AI.

5. AI for Social Good – Beyond the Tech
   AI applications were showcased in healthcare, agriculture, disaster response, and inclusion. Examples included telemedicine, AI-powered food supply chain optimization, and digital agriculture platforms for smallholder farmers.

Recommendations for Stakeholders

1. For Governments and Regulators

• Integrate standards into legislative drafting: Use international standards as scaffolding for national laws, especially in rapidly evolving sectors like generative AI and autonomous systems.

• Invest in inclusive participation: Fund Global South engagement and civil society involvement in SDOs (Standards Development Organizations).

• Establish national AI standards hubs: Mirror the UK’s AI Standards Hub model to coordinate domestic standard-setting with international processes.

2. For Standards Bodies (ITU, ISO, IEC, IEEE, etc.)

• Accelerate update cycles: Abandon decade-long cycles in favor of agile methodologies that reflect AI’s rapid evolution.

• Create ethical assurance frameworks: Develop standards that assess not just performance but ethical behavior and environmental impact of AI systems.

• Strengthen inter-agency collaboration: Expand cross-sector and inter-standard body partnerships to prevent fragmentation.

3. For Tech Companies and Developers

• Adopt standards-by-design: Build systems that comply with emerging standards for explainability, fairness, and energy efficiency from the outset.

• Contribute to open datasets and standards: Share insights and research findings with SDOs to democratize AI governance.

• Engage in skill-building: Ensure AI teams are trained in responsible AI practices, using tools like regulatory sandboxes and ethical checklists.

4. For Civil Society and Academia

• Monitor standardization processes: Act as watchdogs and intermediaries to ensure fairness, transparency, and accountability.

• Facilitate public education: Translate complex AI standards into understandable frameworks for broader public scrutiny.

• Support brainwave donation and inclusive AI: Encourage participation in projects like Inclusive Brains that promote assistive and adaptive technologies for social good.

Conclusion

The AI for Good Global Summit 2025 reveals a critical juncture in global AI governance—one where standards are emerging as the pivotal mechanism to operationalize ethics, ensure interoperability, and enable safe deployment at scale. The summit successfully connected stakeholders from all corners of the ecosystem—governments, industry, academia, civil society—and offered a roadmap to align technical progress with human values.

While challenges remain in inclusivity, geopolitical sovereignty, and the environmental footprint of AI, the launch of the AI Standards Exchange Database, multimedia authenticity frameworks, and sustainability metrics demonstrates that real progress is underway.

The road ahead requires a multistakeholder, multilateral, and multidisciplinary approach—one in which standards are not an afterthought, but the foundation of global AI governance. Without them, the risks of fragmentation, misuse, and inequity will only intensify. With them, AI can truly be a force for good.