Red Line
Responses
In your opinion, what outcomes would make the first Global Dialogue on AI Governance a success?
Thank you, esteemed Co-Chairs. The Global Dialogue will add value only if it produces short-term governance convergence — not language that everyone can sign and no one needs to implement. That means the Co-Chairs will face a real choice. If July tries to cover all seven mandated topics at equal depth, the result will be broad and shallow. The Dialogue will be remembered for its inclusion, not its outcomes. The alternative is not a narrow tent — It is a better structure. Keep participation fully inclusive, but build within that a mechanism for willing actors to go further: voluntary coalitions that demonstrate feasibility on specific issues and report back with implementable outputs. Inclusion and ambition are not in tension if the structure allows for both. I thank the Co-Chairs and the joint Secretariat for their efforts, and I look forward to a substantive Dialogue in Geneva.
From your perspective, which of the following thematic areas identified by the General Assembly Resolution 79/325 for the AI Dialogue reflect your priorities for urgent action and active engagement?
- Transparency, accountability, and human oversight
- Open-source software, open data and open AI models
Please briefly explain your selection.
5
Transparency and accountability for frontier AI systems is the strongest starting point. It is where international cooperation is most needed, most technically tractable, and where existing initiatives - from the Bletchley process to the EU AI Act - have created enough common ground to build on rather than start again. To underline, there is no time to restart from scratch. It means concrete expectations around evaluation reporting, incident reporting, and clear disclosure of system capabilities and limits. One specific gap worth naming: frontier open model release governance. For the most capable systems, a release-first, observe-later model is insufficient. The Dialogue should create space for serious discussion of proportionate ex ante risk assessment for high-risk open model release. This is technically complex and politically contested, which is precisely why a scientific panel and a multilateral forum are the right place to have it. The success of the first Dialogue should be judged by whether it delivers at least one implementable transparency baseline and a pathway for voluntary adoption.