United Nations Association of Orange County (UNA-OC)

A successful Global Dialogue on AI Governance should move beyond broad principles and produce a governance framework capable of addressing how AI redistributes power in society. The first measure of success would be the establishment of minimum human rights–based governance baselines, especially where AI systems affect expression, access to information, equality, privacy, and public participation. These baselines should include transparency, independent oversight, contestability, and access to effective remedy. Second, the Dialogue should recognize that AI systems are not neutral technical tools. In many contexts, they function as infrastructures of visibility, determining what information is amplified, suppressed, prioritized, or made accessible. This has direct implications for freedom of expression and democratic participation. Third, success requires meaningful participation by civil society, affected communities, independent experts, and underrepresented groups, not only as consultees but as actors who can shape standards, oversight mechanisms, and accountability processes. Finally, the Dialogue should produce a practical pathway for international cooperation that reduces regulatory fragmentation while respecting local contexts. The goal should not be uniformity, but alignment around core protections grounded in international human rights law. The Dialogue will be successful if it helps shift AI governance from voluntary commitments and abstract ethics toward enforceable, rights-based, and human-centered accountability.

• Protection and promotion of human rights
• Transparency, accountability, and human oversight
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Interoperability of governance approaches

These priorities reflect the need to address AI governance as a structural human rights issue, not merely a technical or economic question. The protection and promotion of human rights must be central because AI systems increasingly affect how individuals access information, participate in society, seek services, and exercise rights. Without a human rights framework, governance risks focusing only on efficiency or innovation while overlooking harm to dignity, equality, and participation. Transparency, accountability, and human oversight are essential because affected individuals often cannot understand, challenge, or remedy AI-driven outcomes. Transparency should not be limited to general disclosure; it must be meaningful enough to support contestability, independent review, and accountability. The social, ethical, cultural, linguistic, and technical implications of AI are urgent because AI systems are developed and deployed within unequal societies. They can reproduce discrimination, marginalize non-dominant languages and knowledge systems, and reshape cultural and public discourse. Interoperability is also critical because AI systems operate across borders while governance remains fragmented. International cooperation should establish minimum human rights baselines that prevent regulatory gaps and protect individuals regardless of where systems are designed or deployed. Together, these priorities address AI as a system of power that must be governed through law, accountability, inclusion, and human dignity.

One emerging issue is the shift from direct regulation of speech to regulation through digital infrastructure. AI systems increasingly determine what is visible, amplified, removed, or deprioritized. This affects freedom of expression not only through censorship, but through algorithmic control over public attention and access to information. A second issue is design-stage governance. Many human rights impacts are embedded before deployment, through choices about training data, system objectives, optimization criteria, and risk thresholds. Governance must therefore address upstream design decisions, not only post-deployment harm. Third, there is a serious asymmetry of knowledge and power. A small number of actors have the technical capacity to build, audit, and interpret AI systems, while affected individuals and communities often lack the information or resources needed to understand or challenge outcomes. This creates a democratic legitimacy gap. Fourth, AI governance must address the normalization of opacity. If automated systems become widely accepted without explanation, appeal, or remedy, societies may begin to tolerate diminished rights protections as ordinary. These issues require governance that treats transparency, contestability, and remedy as structural requirements, not optional safeguards. AI governance should embed human rights protections throughout the full lifecycle of AI systems: design, deployment, monitoring, and accountability.

In the United States and the broader Western Europe and Other States region, governance gaps are increasingly visible in the everyday use of AI systems that shape information, access, opportunity, and participation. One major challenge is the lack of enforceable accountability for algorithmic systems that curate information and influence public discourse. These systems operate as private infrastructures of visibility, shaping what people see, share, and challenge, yet they often remain outside meaningful transparency and due process standards. A second challenge is regulatory fragmentation. Different sectors and jurisdictions are developing separate approaches, but without a coherent baseline grounded in international human rights standards. This creates uneven protection and risks allowing powerful actors to benefit from regulatory gaps. A third challenge is the limited ability of individuals to contest AI-driven outcomes. Whether in content moderation, public services, employment, education, or access to information, affected individuals often lack clear explanations, appeal mechanisms, or independent review. At the same time, there are important opportunities. Civil society, academic institutions, technical experts, and public-interest organizations are increasingly developing rights-based tools, auditing methods, and policy proposals. These efforts can support stronger oversight if connected to institutional mechanisms. The central opportunity is to move from voluntary principles toward enforceable structures that make AI systems understandable, challengeable, and accountable to the people they affect.

The AI Dialogue can play a critical role by shifting international cooperation from general coordination toward shared accountability. AI systems are developed, deployed, and used across borders, while legal protections remain fragmented. This creates governance gaps, especially when systems designed in one jurisdiction shape rights and opportunities in another. The Dialogue can help establish minimum human rights baselines that apply across contexts, including safeguards for freedom of expression, privacy, non-discrimination, and access to remedy. The Dialogue should also address the concentration of power in AI ecosystems. A limited number of states and private actors currently shape technologies that affect global public life. International cooperation must therefore focus not only on innovation and safety, but on how power is distributed, reviewed, and constrained. Another important role is promoting interoperability among governance approaches. This should not mean imposing one model on all countries. Rather, it should mean aligning around core standards of legality, transparency, accountability, human oversight, and remedy. The Dialogue can also create a space where civil society, affected communities, academia, and technical experts contribute meaningfully to standard-setting and oversight, rather than being included only symbolically. Its added value will depend on whether it helps transform AI governance from fragmented national responses into a coordinated international framework grounded in human rights, public interest, and institutional accountability.

The AI Dialogue should build upon existing initiatives such as the UN Guiding Principles on Business and Human Rights, UNESCO's Recommendation on the Ethics of Artificial Intelligence, the OECD AI Principles, regional regulatory efforts such as the EU AI Act, and emerging work on AI safety, digital rights, and platform accountability. These frameworks provide important foundations, but they remain fragmented. Some emphasize ethics, others safety, innovation, or regulation. Many depend heavily on voluntary compliance or apply only within specific jurisdictions. The result is a gap between principles and implementation. The added value of the AI Dialogue should be to connect these initiatives into a more coherent governance ecosystem. Rather than duplicating existing principles, the Dialogue should focus on operationalizing them through common standards for impact assessment, independent auditing, transparency, human oversight, and access to remedy. The Dialogue can also strengthen cross-regime coherence by connecting human rights law, technical standards, and regulatory practice. This is essential because AI harms often emerge at the intersection of law, technology, markets, and governance. Most importantly, the Dialogue can provide a platform for addressing implementation gaps. Existing frameworks often state what should be protected; the Dialogue should clarify how protection can be enforced, monitored, and challenged in practice. Its strongest contribution would be to move global AI governance from parallel initiatives toward mutually reinforcing systems of accountability.

Different stakeholders should contribute according to their distinct responsibilities and capacities, but participation must move beyond consultation toward influence. Governments should contribute by aligning national regulation with international human rights standards and by supporting enforceable accountability mechanisms. The private sector should provide meaningful transparency, allow independent auditing, and accept responsibility for the societal impacts of AI systems. Civil society should bring rights-based analysis, community perspectives, and independent scrutiny. Academia can support evidence-based research, methodology, and critical evaluation. The technical community should contribute expertise on system design, auditing, safety, and feasibility. The format of the Dialogue should be structured around outcome-oriented engagement. It should include thematic working groups with clear mandates, written submissions followed by deliberative sessions, and transparent reporting on how stakeholder input influences final outcomes. The Dialogue should also include regional consultations to capture different legal, linguistic, cultural, and developmental contexts. Global governance cannot be credible if it only reflects the experiences of technologically dominant actors. Finally, the process should include mechanisms for follow-up. A successful Dialogue should not end with a report; it should create pathways for continued review, implementation, and accountability. The structure should allow stakeholders not only to speak, but to shape the governance agenda.

Global AI governance discussions continue to underrepresent the communities most affected by AI systems. These include marginalized racial, ethnic, linguistic, and religious communities; workers involved in data labeling and content moderation; persons with disabilities; migrants and refugees; children and youth; and communities in regions where AI systems are deployed without strong local regulatory protections. Speakers of non-dominant languages are particularly underrepresented. AI systems often reflect the data, assumptions, and priorities of dominant languages and markets, which can marginalize communities whose knowledge, culture, and expression are less visible in digital systems. Affected individuals are also often excluded because participation requires technical expertise, institutional access, time, language capacity, and resources. This means that those who experience AI harms may have the least ability to influence governance. Inclusion must therefore be intentionally designed. This requires accessible consultation formats, translation, financial support where appropriate, partnerships with trusted local and civil society organizations, and recognition of lived experience as a valid form of expertise. Underrepresented communities should not be included only at the consultation stage. They should have meaningful roles in impact assessment, oversight, auditing priorities, and evaluation of remedies. AI governance will not be legitimate if it is shaped only by those who design systems. It must also be shaped by those who live under their consequences.

The AI Dialogue should use engagement formats that are interactive, evidence-based, and connected to decision-making. One effective format would be case-based deliberation. Participants should examine concrete scenarios, such as algorithmic content moderation, automated welfare decisions, biometric surveillance, or AI-driven misinformation. This allows stakeholders to identify legal, technical, and human rights implications in practical terms. A second format should be structured multi-stakeholder roundtables where governments, companies, civil society, technical experts, and affected communities respond to the same problem from different perspectives. This can reveal conflicts, assumptions, and gaps that general statements often hide. Third, the Dialogue should use iterative feedback rounds. Draft principles or recommendations should be circulated, reviewed, revised, and returned to stakeholders before finalization. This makes participation meaningful and prevents submissions from becoming symbolic. Fourth, regional and community-level consultations should complement global sessions. These formats can capture local realities, linguistic diversity, and unequal capacities that are often invisible in global policy spaces. Finally, digital platforms can be used to collect structured input, map stakeholder priorities, and identify areas of agreement and disagreement. However, digital tools should support deliberation, not replace it. The strongest engagement model would combine expert analysis, lived experience, technical review, and institutional responsibility in a process that produces clear, traceable outcomes.

Effective AI governance requires combining legal, institutional, and technical approaches that translate principles into enforceable protections. Risk-based regulation, such as the EU AI Act, offers one useful model because it links obligations to the level of potential harm. Its value lies in requiring stronger safeguards for systems that affect rights, safety, and public interest. However, risk categories must remain dynamic and responsive to real-world impacts. Algorithmic impact assessments are also important because they move governance upstream. They require actors to examine possible effects on rights, equality, access, and public participation before deployment. To be effective, these assessments should include independent review and public accountability. Independent auditing is another strong practice. Internal compliance alone is insufficient where commercial or political incentives may conflict with human rights protection. Audits should be technically competent, institutionally independent, and linked to remedies. Human rights-based platform governance also offers important lessons, especially in content moderation and information systems. Standards of legality, necessity, proportionality, explanation, appeal, and remedy should guide decisions affecting expression and access to information. The strongest approaches are those that treat accountability as continuous. AI governance should not rely only on post-hoc correction after harm occurs. It should embed human rights safeguards into design, deployment, monitoring, and review. The central lesson is clear: effective AI governance requires systems that are not only innovative, but explainable, contestable, and accountable.