Al Governance Mechanisms With A Human Rights Approach

During the last year, Al governance has proven an urgent priority. As mapped by Kaspar et al. (2023), there is currently more than 50 active international Al governance initiatives, nearly a quarter of them within the UN system itself – including the UN High- level Advisory Body on Artificial Intelligence (HLAB-AI) – joined by other multilateral initiatives or state-led initiatives, including the Council of Europe’s Al Convention, the EU AI Act, and the G7 Hiroshima Process. However, there is a concerning absence of reference to the international human rights law (IHRL) when looking at the different institutional options that have been presented as an alternative for global Al governance. There is a risk that the AI governance discussions sidelines – or worse, neglects to account for – the application of IHRL on Al governance. The authors have conducted a study to assess the human rights impacts of different mechanisms proposed in the public discussion for Al governance and mentioned in the HLAB-AI interim report (2023). The research has collected valuable evidence on what institutional formats and functional elements could provide a strong foundation to ensure the Al global governance mechanisms are underpinned by human rights and support the realisation of the SDGs. This policy briefing leverages the conclusions and recommendations from this study which analyses seventeen governance mechanisms as models for international Al governance. We use this opportunity to summarise lessons and mitigation measures for addressing institutional functions of international Al governance mechanisms, intending to provide insights to guide the design of an international Al governance regime as part of the G20 Framework for Responsible Human Centric Al Governance, but that can also be useful in the engagement of its members states in international Al governance discussions at the UN and other forums.