Cybersecurity at MIT Sloan (CAMS), Sloan School of Management, Massachusetts Institute of Technology

A successful first Global Dialogue on AI Governance would be defined not by symbolic consensus, but by tangible steps toward coordinated, inclusive, and actionable global governance. First, success would mean establishing a shared, interoperable framework for AI governance; one that aligns core principles such as transparency, accountability, and safety across jurisdictions while allowing for contextual flexibility. This would reduce fragmentation and create a foundation for trust and cooperation among nations and stakeholders . Second, meaningful commitments to capacity-building and inclusion would be essential. Many countries and communities remain excluded from shaping and benefiting from AI due to structural barriers in infrastructure, expertise, and access. A successful dialogue would therefore include concrete mechanisms, such as shared infrastructure, knowledge transfer, and funding initiatives, to bridge these divides and ensure broader participation in the global AI ecosystem . Third, the Dialogue should result in the creation of a sustained, multi-stakeholder platform for collaboration. Rather than a one-off event, it should institutionalize ongoing cooperation between governments, industry, academia, and civil society. This platform should support continuous learning, data-sharing, and coordinated responses to emerging risks, reflecting the dynamic and interconnected nature of AI systems. Finally, success would require embedding a systemic and human-centered perspective into governance outcomes. AI challenges, ranging from inequality to security risks and human rights, are deeply interconnected and cannot be addressed in isolation. The Dialogue should therefore promote integrated approaches that align innovation with equity, trust, and societal resilience. In essence, the Dialogue would be successful if it transforms global intent into structured collaboration, bridging divides while building a coherent, trusted foundation for the future of AI governance. This answer is backed by scientific contribution together with O.Alhaddad. A summery will be emailed separately.

• Interoperability of governance approaches
• Safe, secure and trustworthy AI
• AI capacity-building
• Protection and promotion of human rights

These four priorities reflect the most critical and interdependent areas for ensuring that AI development remains stable, inclusive, and aligned with global societal goals. Safe, secure and trustworthy AI is essential due to the increasing complexity and risk profile of AI systems. As AI capabilities expand, so do vulnerabilities, including misuse, adversarial threats, and systemic security risks. Addressing these challenges requires a socio-technical approach that goes beyond technical safeguards to include governance, organizational practices, and broader societal resilience . AI capacity-building is fundamental to reducing global inequalities in AI development and adoption. Many countries and communities lack access to infrastructure, expertise, and data, limiting both their ability to benefit from AI and to shape its governance. Targeted investments in education, infrastructure, and institutional capacity are therefore critical to enable equitable participation and to strengthen the overall quality and diversity of AI systems . Protection and promotion of human rights must be embedded at the core of AI governance. AI systems can amplify bias, enable surveillance, and concentrate power, posing risks to fairness, autonomy, and dignity. Effective governance requires operationalizing human rights through enforceable mechanisms such as transparency, accountability, and continuous oversight across the AI lifecycle . Finally, interoperability of governance approaches is crucial to address fragmentation across jurisdictions. Divergent regulatory frameworks risk creating gaps, inconsistencies, and inefficiencies that undermine trust and innovation. Greater alignment through shared standards, mutual recognition, and coordinated international efforts can support more coherent and effective global AI governance. Together, these priorities reinforce a systemic approach where inclusion, trust, and coordination act as stabilizing mechanisms for responsible AI adoption. This answer is backed by scientific contribution together with O.Alhaddad. A summery will be emailed separately.

While the listed themes capture many core dimensions of AI governance, several cross-cutting and emerging issues remain insufficiently explicit and warrant dedicated attention. First, the systemic nature of AI adoption itself is underrepresented. AI-related challenges ,e.g. spanning inequality, security, governance, and human rights, do not operate in isolation but evolve through interconnected feedback loops. Policies that address these domains separately risk unintended consequences elsewhere. A more explicit focus on systems thinking and integrated governance approaches is needed to manage these interdependencies effectively . Second, the emergence of AI-driven socio-economic instability, particularly linked to workforce transitions, deserves greater prominence. Automation can create displacement, skill mismatches, and social exclusion, which may in turn generate new forms of risk, including insider threats and declining trust in institutions. These dynamics highlight the need to connect AI governance more closely with labor, education, and social policy frameworks . Third, trust as a dynamic and measurable governance objective is not sufficiently emphasized. Trust is not only an outcome but also a prerequisite for effective AI systems, influencing participation, data quality, and adoption. Declining trust can reinforce exclusion and reduce system performance, suggesting the need for continuous monitoring and trust-building mechanisms across the AI lifecycle . Finally, the concept of global stabilizing mechanisms, such as shared infrastructure, coordinated monitoring systems, and inclusive governance architectures, remains underdeveloped. Current approaches are often reactive and fragmented, while the scale and speed of AI development require proactive, globally coordinated responses. Addressing these cross-cutting issues would strengthen the Dialogue by ensuring that AI governance frameworks are not only comprehensive, but also adaptive, integrated, and resilient in the face of evolving systemic risks. This answer is backed by scientific contribution together with O.Alhaddad. A summery will be emailed separately.

Governance gaps across the selected thematic areas are increasingly global in nature, requiring coordinated action among countries and regions rather than isolated national approaches. A central challenge is the fragmentation of governance approaches across jurisdictions. Divergent regulatory frameworks, standards, and enforcement mechanisms create uncertainty for organizations operating globally and may lead to regulatory arbitrage or uneven risk management. This fragmentation can slow innovation and weaken trust. At the same time, it presents a major opportunity: through international collaboration, countries can work toward interoperable frameworks and shared principles that enable both innovation and consistency at a global scale . In the domain of safe, secure, and trustworthy AI, risks are inherently transnational. Cyber threats, misuse of AI, and systemic vulnerabilities do not respect borders, and weaknesses in one region can have cascading global effects. This underscores the need for collective security approaches, shared risk intelligence, and coordinated safeguards. Such collaboration also creates opportunities to build globally trusted AI ecosystems grounded in resilience and mutual accountability . Capacity-building gaps remain one of the most significant global challenges. Unequal access to infrastructure, data, and expertise limits the ability of many countries to participate in and shape AI development. Without coordinated international support, these disparities risk deepening global inequality. However, joint initiatives, such as shared infrastructure, knowledge exchange, and funding mechanisms, can enable more inclusive participation and strengthen the overall global system . Finally, human rights considerations require global alignment. AI systems can amplify bias and enable harmful practices across borders, making it essential to move beyond principles toward enforceable, internationally recognized safeguards. Overall, addressing these governance gaps collectively offers an opportunity to build a more coherent, inclusive, and trusted global AI governance ecosystem. This answer is backed by scientific contribution together with O.Alhaddad. A summery will be emailed separately.

The AI Dialogue can play a pivotal role as a global convening and coordination platform, enabling countries, industries, and stakeholders to move from fragmented efforts toward more coherent and cooperative AI governance. First, it can foster alignment on shared principles and interoperable frameworks. By bringing together diverse perspectives, the Dialogue can help translate high-level values, such as safety, accountability, and human rights, into more consistent and compatible governance approaches across jurisdictions. This reduces fragmentation and creates a foundation for trust and cross-border innovation . Second, the Dialogue can act as a catalyst for collective capacity-building. Many countries lack the resources and expertise to fully participate in AI development and governance. Through coordinated initiatives, such as knowledge-sharing, technical assistance, and shared infrastructure, the Dialogue can help bridge these gaps and promote more equitable global participation . Third, it can support ongoing multi-stakeholder collaboration. AI governance is not a one-time effort but an evolving process. The Dialogue can institutionalize continuous engagement between governments, industry, academia, and civil society, ensuring that policies remain adaptive and grounded in real-world developments. This includes facilitating the exchange of best practices, data, and lessons learned. Finally, the Dialogue can promote a systemic and integrated approach to AI governance. By recognizing the interconnections between economic, social, security, and human rights dimensions, it can help stakeholders address risks and opportunities holistically rather than in isolation. In essence, the AI Dialogue can serve not only as a forum for discussion, but as an enabling mechanism for sustained international cooperation, aligning global efforts to ensure that AI development remains safe, inclusive, and beneficial for all. This answer is backed by scientific contribution together with O.Alhaddad. A summery will be emailed separately.

Several MIT-based initiatives provide strong foundations for an AI Dialogue, particularly those integrating AI, governance, and cybersecurity across disciplines. A central initiative is Cybersecurity at MIT Sloan (CAMS), which functions as a cross-sector consortium bringing together researchers, CISOs, and executives to address cyber risk at the intersection of AI, governance, and systemic resilience . CAMS' work is especially relevant because it reframes cyber risk as a board-level governance challenge, emphasizing judgment, institutional resilience, and interconnected system risks rather than purely technical metrics . Its research on AI-driven cyber threats, global governance gaps (e.g., UN cybercrime frameworks), and systemic risk propagation directly aligns with the needs of an international AI Dialogue . CAMS actively contributes to AI & Cybersecurity Strategic Dialogues that help shape the future research roadmap for AI governance across MIT's various labs and research entities. This engagement positions CAMS uniquely from a business school perspective, bridging strategy, cyber risk governance, and technology deployment while offering an integrated, holistic lens that is often missing in purely technical AI initiatives. In doing so, CAMS extends its impact beyond its own domain, strengthening AI research across MIT's broader ecosystem, including MIT Computer Science and Artificial Intelligence Laboratory and the MIT Schwarzman College of Computing. While these entities provide deep technical expertise, CAMS adds distinct value by translating and embedding these insights into governance, financial risk, and organizational decision-making frameworks. Added value for an AI Dialogue: Building on MIT CAMS would enable (1) integration of AI risk into enterprise and policy governance, (2) cross-sector dialogue between academia, industry, and regulators, and (3) a systemic, interdisciplinary approach linking AI safety, cybersecurity, economics, and institutional design. Leveraging MIT's interconnected research model ensures the Dialogue is not siloed, but instead grounded in holistic, real-world decision contexts across all AI disciplines.

Global AI governance debates remain dominated by governments, large tech firms, and well-resourced research institutions. This creates structural blind spots, as participation tends to mirror existing inequalities in access, data, and infrastructure . Underrepresented voices include: 1. Low- and middle-income countries (LMICs): Limited infrastructure and expertise restrict their influence, despite being heavily impacted by AI-driven economic and social shifts. 2. Marginalized communities (e.g., women, ethnic minorities): AI systems often reproduce existing biases, yet those affected are rarely involved in design or governance. 3. Workers and displaced labor groups: Automation reshapes labor markets, but affected workers have minimal input into governance despite facing direct consequences. 4. Civil society and grassroots organizations: Particularly in the Global South, these actors lack access to global forums where standards are set. 5. Human rights defenders and activists: Especially those targeted by AI-enabled harms (e.g., surveillance, deepfakes), yet excluded from policy design. How to include them: 1. Capacity-building: Invest in education, infrastructure, and local AI ecosystems to enable meaningful participation from under-resourced regions. 2. Inclusive governance mechanisms: Move beyond symbolic consultation to co-creation models where affected groups shape agendas and outcomes. 3. Shared infrastructure and open systems: Open-source tools and data can lower barriers to entry and diversify contributors. 4. Institutionalized representation: Reserve seats or formal roles for marginalized groups in global forums (e.g., UN-led processes). 5. Human rights integration: Embed lived experiences into oversight frameworks, ensuring governance reflects real-world harms and contexts. Overall, inclusion is not just normative, it is a functional requirement for effective and legitimate AI governance, improving trust, data diversity, and system resilience. This answer is derived from scientific contribution together with O.Alhaddad. A summery will be emailed separately.

To move beyond static, top-down discussions, AI governance dialogues should adopt formats that prioritize interaction, co-creation, and real-world grounding. 1. Multi-stakeholder co-creation labs. Small, diverse groups (policy, industry, civil society, affected communities) work on concrete challenges (e.g., bias mitigation, AI safety). Outputs are (the basis for) draft policy tools or prototypes, not just discussion. This shifts participation from symbolic to substantive. 2. Scenario-based simulations ("policy sandboxes"). Participants engage in role-play around realistic AI crises (e.g., deepfake election interference, labor displacement shocks). This helps (in an interactive and close to real-world experience) uncover trade-offs across security, ethics, and economics, reflecting the interconnected nature of AI systems . 3. Lived-experience panels with deliberative forums. Instead of expert-only panels, pair affected individuals (e.g., workers, activists) with policymakers in moderated dialogues. Follow with small-group deliberation sessions to translate experiences into governance recommendations. 4. Iterative "dialogue loops". Replace one-off events with staged engagement (i) Phase 1: Input gathering (global, digital, inclusive), (ii) Phase 2: Synthesis workshops, (iii) Phase 3: Feedback validation with original contributors. This ensures continuity, trust, and accountability. 5. Open digital participation platforms. Use multilingual, accessible platforms for crowdsourcing ideas, voting on priorities, and commenting on draft frameworks, broadening participation beyond those physically present. 6. Cross-cluster integration sessions. Rather than siloed themes (e.g., safety vs. inclusion), host facilitated sessions explicitly designed to map interdependencies and tensions, aligning with the need for systemic thinking in AI governance . 7. "Reverse panels". Decision-makers listen while community representatives ask questions and set the agenda, rebalancing power dynamics. Together, these formats foster deeper engagement by making dialogue participatory, continuous, and action-oriented, turning discussion into shared problem-solving.