Can ICANN's Multistakeholder Model Serve as a Blueprint for AI Governance?
Mohamed (Mo) Elbashir
Infra Enablement | AI Governance & Tech Policy | GeoTech | Geopolitics & Regulatory | Empowering Infra-Tech Innovators @Meta | Senior Fellow Atlantic Council GeoTech Center
In a recent Washington Post op-ed titled "Who Will Control the Future of AI?," OpenAI CEO Sam Altman proposed using the Internet Corporation for Assigned Names and Numbers (ICANN) multistakeholder model as a template for global AI governance. This is a shift from previous AI governance proposals, which frequently drew comparisons between AI risks and biological or nuclear control regimes. As a long-time active participant in ICANN's community and Internet governance processes, I believe the suggestion deserves careful consideration and study. Given ICANN's two-decade track record of managing critical Internet infrastructure stability and growth, its multistakeholder approach provides valuable insights and a potential path forward for inclusive multistakeholder AI governance.
ICANN's Multistakeholder Model: Overview
ICANN is a non-profit organization that manages the global domain name system and IP address allocations. It was founded in 1998 with support from the United States government, incorporated in California,?and has since become?an independent global organization. ICANN convenes governments, technology companies, civil society, and technical experts in a consensus-driven, bottom-up process that ensures broad participation and transparency.
ICANN's primary goal?is to ensure the global Internet's stability, security, and openness. It encourages collaboration among its stakeholders (Supporting Organizations and Advisory Committees) to address Internet governance issues through open and decentralized decision-making. While imperfect, this model has effectively prevented Internet fragmentation and maintained global Internet stability.
Challenges of ICANN's Journey
Despite its success, ICANN has dealt with significant geopolitical challenges. Some governments, particularly Russia and China, questioned ICANN's legitimacy, claiming that the United States maintained undue influence over global internet policy. This prompted calls for more government-led management of the Internet's critical resources (domain names and IP addresses) or establishing alternative Internet governance frameworks, posing the threat of Internet fragmentation or "balkanization."
The 2016 IANA function stewardship transition marked a watershed moment in ICANN's history. The United States has relinquished its oversight role, allowing ICANN to transition to a?true global multistakeholder governance organization. This transition demonstrated the practical feasibility of a bottom-up, consensus-based governance model for critical Internet resources. As co-chair of the IANA coordination working group, I was honored to work with 30 diverse?Internet experts?involved in the development of the transition proposal; they represented?stakeholders such as the Internet Engineering Task Force (IETF), International Chambers of Commerce (ICC), Internet Society, government representatives from the United States, Iran, Brazil, Egypt, India, and China, as well as tech companies such as Yahoo and Facebook.
Adapting the ICANN Model for AI Governance
To determine whether ICANN's multistakeholder model suits?AI governance, we must compare its strengths and limitations to key characteristics such as inclusivity, transparency, accountability, legitimacy, and sustainability.
Challenges of Establishing an ICANN-like AI Governance Body
There is an ongoing global effort to propose and develop a global AI governance framework led by The UN Secretary-General's AI Advisory Body, which is expected before the end of 2024. In its interim report, "Governing AI for Humanity," the advisory body called for universal, networked, and multistakeholder AI governance that should prioritize universal buy-in by countries and stakeholders. It recommended that the governance regime leverage "existing institutions" through a networked approach.
领英推荐
Historical factors have led to the current dominance of governments in policy development and engagement at the UN and its intergovernmental organizations. Although they are significant participants and decision-makers, they are also stakeholders in influencing AI's development.?Other stakeholders include the tech community, civil society, academia, and, most importantly, AI users whose voices should be heard. A trustworthy multistakeholder process should facilitate the inclusion of all participants and ensure their contributions are considered during decision-making, similar to ICANN's multistakeholder model. Developing such a global model is one of the most difficult challenges of establishing global inclusive AI governance.
Given AI's far-reaching implications for national security and socioeconomic impact, governments are unlikely to accept or give control to a global multistakeholder AI governance body. Unlike the Internet, where the United States government played a sole role in ICANN's formation, AI governance will need more complex diplomacy and broad government buy-in and participation to ensure legitimacy and acceptance.
While determining the best AI governance decision-making body is critical, the United States and other democratic governments (e.g., the United Kingdom, Japan, South Korea, and Singapore) could propose and advocate for a new multistakeholder AI governance body and progress towards this goal.
Potential Advantages of Adopting ICANN's Model for AI Governance
Despite historical and technological differences in Internet and AI development, some aspects of ICANN's multistakeholder approach may benefit AI governance:
Specialized Working Groups: AI governance policies are developed through focused groups on ethics, safety, and societal impact.
Advisory Function: Allow AI experts and the general public to participate in consultations, ensuring widespread participation and increasing expert influence in governance decisions.
Ethical Framework: Develop shared ethical principles for AI development and deployment through multistakeholder policy development.
AI Regulation Coordination: Encourage international collaboration between national regulatory bodies to standardize AI governance frameworks.
Conclusion
ICANN's multistakeholder model provides valuable insights into AI governance. Its inclusive, transparent, and global approach could contribute to a more ethical and responsible AI landscape. However, the direct application of ICANN's model to AI necessitates addressing specific AI-related issues such as rapid innovation, concentrated power dynamics, and ethical complexity. The future of AI governance will likely require combining ICANN's multistakeholder consensus-driven principles with more proactive, responsive risk mitigation mechanisms tailored to AI's specific needs.
"The views expressed in this article are my own opinions and not those of my employer or my affiliations."
References:
Professor of Chemistry; Information Society, Internet Governance, e-learning, e-government. Internet Hall of Fame class of 2021.
6 个月There is precedent to this, see my article from 2021 in https://pisanty.blogspot.com/2021/06/federated-platform-governance-step.html
Data Analytics Expert | AI Specialist | Market Research Leader
6 个月“History never repeats itself, but it does often rhyme.” We could do worse than to allow the ICANN multistakeholder model to inform an AI governance model.
Interesting point of view. As Avri Doria mentions, each multistakeholder organization is different. A basic question is what is the goal and what are the functions of the organization (form follows function as Jovan Kurbalija would say). At the same time the multistakeholder approach is not ?n‘importe quoi“, but should have some substance in order to avoid any ?multistakeholder-washing“. A key reference in that regard are the S?o Paulo Multistakeholder Guidelines adopted recently in Brazil at the NetMundial+10 - check them out under Section 3.2 of the Final Statement https://netmundial.br/pdf/NETmundial10-MultistakeholderStatement-2024.pdf
Board member, Managing Director, Standards Author, Industry Speaker, Head of Information Security and Engineering Departments
6 个月I would caution folks on using the ICANN model as a blueprint, certainly learn from the positives and many negatives of ICANN and ensure that a governance structure for AI undertands and considers the data source regulions and risks inherent in the LLM. Don't fall into a trap thet ICANN is shiny because it's not government - that in itself is a false narrative. Also don't assume that every thing in the multi stakeholder model is amateurish. Any governance structure is primarily seated with the influential constituents and wat they see as risk and reward.
Business Analysis | Artificial Intelligence | Entrepreneurship
6 个月Adopting a multistakeholder model similar to ICANN for AI governance is a powerful idea, as it builds on extensive experience in managing complex global systems. However, this approach would require a shift in priorities—from focusing on interoperability and technical management to emphasizing ethical governance and the safety of humanity. ICANN and similar organizations have gained influence largely due to the direct value they offer through ensuring interoperability. In the case of AI, the challenge will be even greater, as governance must balance technological progress with ethical considerations, particularly in a field known for intense competition, sometimes sponsored by states.