Cogitait

The Global Digital Compact will be a success if it moves beyond declarations and delivers concrete outcomes. First, governance frameworks that are practical and enforceable: grounded in the four priorities I identified: safety, human rights, transparency, and an understanding of AI's broader social and economic implications. Second, a multi-stakeholder process that is genuinely structured: where private sector input, particularly from those building and deploying AI systems, has a clear and visible connection to policy decisions, not just side events. Third, a certification process for public sector AI use : a shared standard developed with industry from the start, that governments can apply when procuring or deploying AI. That would be a tangible, lasting output from this Dialogue. A certification for the private sector to work with public sector would be advisable. Success is a set of tools, standards, and partnerships that governments and industry can actually use.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

Across all four, the common thread is the need for governance that is not just well intentioned, but concrete and implementable. Developed with input from those who build, deploy and are affected by these systems.

Informing and protecting AI consumers

In the AI development and deployment sector, governance gaps are having a direct and practical impact on how we build and bring systems to market. The most pressing challenge is the exclusion of the private sector from meaningful policy decisions. Regulations are increasingly being developed without structured input from those who actually build these systems, resulting in frameworks that are well-intentioned but difficult to implement. Without shared global standards, companies are navigating contradictory national and regional requirements. Without industry involvement in designing accountability frameworks, the mechanisms being proposed are not always technically feasible. There is also a broader challenge: public trust. AI is still widely perceived as a threat rather than a tool and that fear reflects a real absence of visible accountability. This is why we propose a certification process for AI companies, a recognisable standard signalling that a company is building and deploying AI responsibly. It would serve two purposes: helping companies follow the right governance frameworks and giving governments and the public a clear signal of trustworthiness. That is exactly the kind of concrete, practical output this Dialogue could deliver.

The AI Dialogue has a unique opportunity to do something that national and regional processes cannot, create a truly global foundation for AI governance that works for everyone. From a private sector perspective, there are three concrete ways the Dialogue can advance international cooperation. 1- by bridging the gap between policy and practice. International cooperation is only meaningful if it produces frameworks that are actually implementable. That requires structured private sector involvement from the start, not consultation at the end. 2 - by establishing shared standards. A certification process for AI companies — developed through the Dialogue with input from governments, industry, and civil society, could become a globally recognised benchmark for responsible AI development. This would reduce regulatory fragmentation and build public trust across borders. 3- by making trust a shared goal. AI governance is not just a technical or legal challenge — it is a public confidence challenge. The Dialogue can model what responsible, inclusive and transparent cooperation looks like. In doing so, demonstrate that AI can be governed well.

There are several existing initiatives the AI Dialogue should actively build upon rather than duplicate. The OECD AI Principles and the EU AI Act have already established important foundations around safety, transparency and accountability. The Dialogue should use these as reference points while ensuring they are accessible and applicable beyond high-income countries and regions. The UNESCO Recommendation on the Ethics of AI brings a human rights and cultural diversity lens that is essential for a truly global framework and aligns directly with the thematic priorities I identified. The ITU's AI for Good platform and existing ISO/IEC technical standards provide practical infrastructure that the Dialogue can connect governance commitments to real implementation. The added value the AI Dialogue can bring is what none of these initiatives have fully achieved: a single inclusive space where governments, industry, and civil society align around common standards, including a certification process for AI companies that draws on all of the above and translates them into a recognisable, practical benchmark. The Dialogue does not need to start from scratch. It needs to connect the dots and deliver something concrete that the world can use.

Effective contribution from different stakeholders requires a deliberately designed structure. Governments bring regulatory authority and political legitimacy. Their role is to set the boundaries and make binding commitments. But they are most effective when informed by those closest to the technology. The private sector brings deployment experience, technical knowledge and infrastructure. However, as I have highlighted throughout, this contribution is only meaningful if it is structured. We recommend designated industry input tracks with clear feedback loops into policy deliberations, not parallel side events. A practical starting point would be involving industry in the design of a certification process for AI companies, where our expertise is directly relevant. Civil society and academia provide the independent scrutiny and long-term thinking that keeps the process honest and grounded in real-world impact. On format, we recommend moving away from traditional plenary-heavy structures toward smaller, thematic working groups that produce tangible outputs. Each group should have a clear mandate, mixed stakeholder composition, and a defined deliverable, whether a standard, a framework, or a recommendation. The Dialogue should be measured not by the number of statements delivered, but by the number of concrete agreements reached.

Global AI governance discussions remain heavily skewed toward high-income countries, large technology companies and English speaking institutions. Several critical voices are consistently underrepresented. Small and medium-sized AI companies are often absent from these conversations despite being at the forefront of deployment. Large corporations have the resources to engage in international policy processes; smaller ones do not. The certification process we have proposed could serve as a natural entry point to bring this community in. The Global South remains underrepresented both in the rooms where decisions are made and in the datasets and systems that AI governance is meant to regulate. This is an effectiveness issue. Governance frameworks built without this input will simply not work globally. Women and underrepresented groups in tech bring perspectives that are essential for identifying bias, exclusion, and harm in AI systems. Initiatives like UNESCO's Women4Ethical AI are a step in the right direction and should be formally connected to the Dialogue's work. On inclusion, the solution is structural. Reserved seats, dedicated funding for participation, multilingual engagement, and regional preparatory processes are not optional extras, they are what makes the Dialogue legitimate.

The AI Dialogue should deliberately experiment with more dynamic engagement formats. Thematic working groups with mixed stakeholder composition should replace or complement large plenaries. Each group should have a specific mandate, a time limit and a required output, whether a draft standard, a shared definition or a concrete recommendation. This is where real progress happens. Problem-solving sessions framed around real world scenarios rather than abstract principles would help bridge the gap between policy and practice. Bringing a live governance challenge to the table, for example, how to certify an AI system for public sector procurement, forces participants to engage concretely rather than rhetorically. Structured private sector roundtables with direct reporting into the main Dialogue would ensure industry input is visible and traceable. Digital participation mechanisms should be designed for genuine contribution, not just observation. Asynchronous input tools, multilingual platforms and regional satellite sessions would allow voices from the Global South and smaller organisations to engage meaningfully without the barrier of travel costs. Finally, every session should end with a visible output, even a short one.

Please refer to my previous answers. I have concrete examples there of formats, policies and approaches.