Association for Progressive Communications (APC)

-Inclusive and accountable multistakeholder model of AI governance: The Dialogue should reflect a genuine multistakeholder approach, identifying practical ways to finance and build capacity in Global South countries, including support for local research to inform evidence-based policy making. It must ensure meaningful participation of stakeholders – especially from the Global South – throughout the process, following the São Paulo Guidelines for inclusivity, transparency, and accountability. It should also serve as an agenda-setting platform that connects with the Commission on Science and Technology for Development (CSTD) Working Group on Data Governance to avoid fragmentation in developing rights-based data governance principles. - Commitment to human rights-based approach to AI that is rooted in international human rights law: As outlined in paragraph 55 of the Global Digital Compact (GDC), the Dialogue and its outcomes must centre human rights and be grounded in international law. It should define human rights-based AI, clearly identify risks and harms, and adopt a precautionary approach – including moratoria on rights-violating systems, thresholds for mandatory due diligence, and private sector compliance with business and human rights standards. - Address digital colonialism in the design, development and deployment of AI: The Dialogue must address the concentration of AI design, development and deployment in a few Global North countries and corporations, which risks reinforcing inequalities and creating new forms of digital colonialism. It should promote resource and capacity transfer to ensure Global South participation – especially from historically excluded groups – in AI leadership and decision making. - Mainstream gender as a cross-cutting issue across the Dialogue: The Dialogue must make gender equality a core principle of AI governance, addressing underlying power structures both offline and online. Gender should be mainstreamed across all areas, with AI's gendered risks identified and addressed through robust analysis, mandatory impact assessments, and gender-just data governance principles.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

The UN GDC's commitment to respecting international human rights law across the entire AI lifecycle must be fully honoured. Priority should be given to the real and emerging harms already affecting many countries, particularly in the Global South. These include exploitation in data work, inadequate labour protections, unequal access to education and digital infrastructure, technology-facilitated gender-based violence, and risks to safety and security in real-world AI deployments. Addressing these issues demands that AI governance be rooted in lived experiences, not speculative future existential threats. Further, AI is not merely a technological innovation, but a social, economic and cultural infrastructure that can either democratise or colonise the global commons. It is therefore critical that the knowledge, language and realities of its users are embedded in its design, through ensuring their equitable representation across the AI value chain, particularly from the Global South. All AI governance efforts at all levels must be firmly anchored in human rights. This entails developing recommendations on prohibited practices and applications, including standards to implement moratoria on high-risk systems - such as the use of AI in combat, military or war situations - as called for by the Office of the United Nations High Commissioner for Human Rights (A/HRC/48/31), as well as guidance on human rights due diligence and risk mitigation. Together, these steps would allow the Dialogue to move beyond a limited scope and address the full range of AI-related impacts and harms. An immediate move toward a precautionary approach in developing and deploying AI is essential. This calls for an independent mechanism to (a) require mandatory ex ante risk assessments for high-risk systems, and (b) ensure accountability - both corporate and public - across the AI supply chain through post-deployment audits.

- AI and sustainable development: AI innovation requires an urgent shift to fully acknowledge and address its human and environmental impacts across the AI value chain. The Dialogue should examine ways to confront the ecological harm, labour exploitation, capital concentration and knowledge extraction embedded in today's AI models. AI governance is inseparable from broader systems of global governance aimed at achieving structural equity and digital justice. - AI commons: The Dialogue should set parameters to enable conditions for democratising the innovation dividends and benefits of the AI paradigm. These include public incentives, including investments in open source communities, social licensing frameworks for data and AI innovation, and community-led data stewardship approaches and models. - Gender equity and online violence: AI regulation must address gender-related harms, especially technology-facilitated gender-based violence (TFGBV). Legal gaps and fragmented approaches leave victims - particularly in cases of non-consensual intimate images (NCII) - without protection or recourse. Stronger laws, audits, indicators and gender-responsive safeguards are needed, alongside requirements for safety, privacy and consent in AI design. Governance frameworks should include gender-based analysis and enforce rules on deepfakes and platform accountability. More broadly, the underrepresentation and digital exclusion of women and gender-diverse people in the development of technology and AI, and in access and governance - especially from the Global South - must be addressed. In line with SDG 5, gender equality should be integrated into the Dialogue from the outset.

Weak or inconsistent compliance with international human rights law across AI design, development and deployment is a key gap. Human rights risk assessment and mitigation must apply throughout the AI lifecycle, including due diligence, ex ante and post hoc assessments, and prohibitions on non-compliant technologies. In many Global South contexts, data extraction and persistent under- or misrepresentation lead to biased and unequal AI outcomes, compounded by limited inclusion in model design. This highlights the need for meaningful community participation in AI governance as well as mandated data audits. These inequities are reinforced by labour exploitation across AI value chains and by the environmental burdens of resource-intensive systems, which disproportionately affect the Global South. Fragmented governance further limits inclusive participation, especially for communities facing resource and language barriers, making greater interoperability essential for more equitable engagement and shaping fairer, pluralistic AI governance models. Key barriers also stem from industry-driven AI risk debates lacking grounding in binding human rights standards. Developing countries face extractive digital trade pressures, unequal IP regimes, lack of global digital taxation, and dominance by large tech firms, all of which constrain autonomy. Limited public financing hampers capacity building, while AI is often framed narrowly as a tool rather than a driver of structural change. Meanwhile, dominant "Big AI" narratives overshadow local innovation and open source development. A meaningful breakthrough requires centring people and public interest in the Global Dialogue, guided by human rights, sustainable development, and structural justice – supporting fair AI models that prioritise human well-being, ecological sustainability, and multistakeholder models for participatory approaches in AI governance.

The Global Dialogue on AI can play an essential role in protecting and promoting human rights and advancing global cooperation for public accountability and corporate liability frameworks in compliance with international human rights law, as a basis for democratising the benefits of AI and protecting against its harms. The Global Dialogue is also key to creating regulatory coherence by integrating data and AI governance based on data justice frameworks, acknowledging the potential harms of AI to people and the planet and the role of collective/community stewardship of data resources. The Dialogue should promote cooperation models that are context-sensitive and globally coherent in order to respond to particular realities while ensuring alignment around common basic safeguards, in accordance with international human rights law. The Dialogue can ensure that Global South states, and underrepresented communities from the Global South, have a proportionate stake in development and governance of AI. For this, multistakeholder participation should be effectively enabled in AI governance, on equal footing, to ensure responsible, secure, transparent and human rights-based AI and AI governance. The AI Impact Summit in New Delhi, India (2026) and the declaration from the summit (https://www.mea.gov.in/bilateral-documents.htm?dtl/40809) marks a missed opportunity to balance the expectations of economic growth and innovation in the field of AI with assurances that such growth will not be at the expense of people, and to ensure that human rights standards are the cornerstone of the development of this technology.

The Dialogue should build on and be closely aligned with existing efforts such as the ones led by the Office of the United Nations High Commissioner for Human Rights, the Human Rights Council, and UN human rights mechanisms such as Special Procedures and treaty bodies. It should also connect with frameworks established through the World Summit on the Information Society (WSIS) and the Global Digital Compact. For example, the CSTD Working Group on Data Governance is particularly relevant, given that data governance underpins AI systems and must reflect human rights standards. The Dialogue should also engage with multistakeholder platforms developed under WSIS, notably the Internet Governance Forum and its network of National and Regional Initiatives. The global IGF scheduled for December 2026 – falling between the two Dialogue sessions – offers a timely opportunity to assess the first Dialogue's outcomes and prepare for the second. The Dialogue should further coordinate with ongoing UN agency work through the UNGIS, particularly the WSIS+20 Internal Task Force on Financing Mechanisms. In addition, insights from NETmundial and NETmundial+10 should help inform the Dialogue's structure and approach. The Dialogue should also draw on established normative instruments such as the Council of Europe's Framework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law, the UNESCO Recommendation on the Ethics of AI, UN Guiding Principles on Business and Human Rights, and the BRICS Leaders' Statement on the Global Governance of AI.

Non-state stakeholders, such as civil society, need to have the opportunity, following modalities of other UN processes, to speak in plenaries, deliver interventions, and participate in dedicated expert panels/briefings. These mechanisms facilitate real-time exchange and dialogue between member states and other stakeholders. The process should allow for meaningful participation by CSOs through written and verbal submissions according to a publicly available timeline of the process of the AI Dialogue. Participation should be open and accessible, taking into account issues around language and access, ensuring hybrid modalities and financing for support and travel for participating in in-person meetings around the Dialogue. The learnings from the WSIS+20 process have indicated that multiple stakeholders can be part of a global governance process, but we also see that the impact of CSOs and their input is often limited while inputs of governments and private companies are given priority. This could be addressed through a multistakeholder advisory group with adequate civil society representation that has a formal role in negotiations, as well as by prioritising interventions by civil society and underrepresented groups in sequencing, so that their agendas can be a central point of discussions.

Women and gender-diverse people – especially from the Global South – remain marginalised and excluded from discussions on AI governance, thereby further marginalising their concerns around safety, security and privacy in relation to AI. A proactive approach is needed, similar to the IGF, by integrating gender-related questions across all calls for submissions and session proposals and ensuring representation in all panels in events related to the Global Dialogue. The use of AI in surveillance, military and war efforts, autonomous weapons and tracking is not central to the debate around the use of AI. This is more possible when those directly impacted by aggression and war in Asia and Africa are proactively included in the discussions on AI governance. Indigenous communities, whose lives and communities are severely impacted by the extractive practices of AI companies through demand for water, critical minerals and energy from their lands, also need to be provided a seat at the table. It is important that these underrepresented groups not only have the opportunity to engage in these discussions, but also play an active role in agenda setting and decision making at all levels of AI governance. To ensure this, building capacities is essential, by funding and supporting experts from such communities, to participate in global standards-setting bodies as well as providing leadership training to build skills in auditing AI systems, designing evaluation frameworks, setting procurement standards, and leading governance processes.

The modalities through which the NETmundial+10 initiative was conducted and the NETmundial+10 Multistakeholder Guidelines serve as positive inspiration for ensuring balanced engagement of governmental and non-governmental actors in the Dialogue, and meaningful, inclusive engagement. A minimum of 50% rotating time must be allocated for non-governmental stakeholders, particularly civil society groups, to engage in dialogue with each other and with governments. The NETmundial+10 Multistakeholder Guidelines can be also a valuable tool for improved coordination and information sharing between the Dialogue and other multistakeholder and multilateral digital governance processes. The Dialogue can also build on and strengthen existing practice from other UN processes, such as the WSIS+20. We recommend the establishment of a Multistakeholder Advisory Group (MAG) to facilitate dialogue and stakeholder engagement throughout the process and opportunities to weigh in on key outcomes for all stakeholders, building on the experiences of the IGF MAG and the Informal Multistakeholder Sounding Board in the WSIS+20 review. We highlight the "Joint letter to the UN Global Dialogue on Artificial Intelligence co-chairs on multistakeholder participation" (https://www.apc.org/en/pubs/joint-letter-un-global-dialogue-artificial-intelligence-co-chairs-multistakeholder) as a set of practical recommendations to put in practice the principles of inclusivity, transparency and meaningful input throughout the Dialogue. In terms of the focus, the Dialogue should address both broad/cross-cutting AI governance themes – such as normative frameworks, human rights, gender, ethics and regulatory alignment – and targeted, sector-specific analyses of AI's risks, impacts and opportunities across fields like healthcare, education, labour, finance, climate and biodiversity, corporate accountability and security. Consider linking Thematic Discussions with intersessional Working Groups to ensure follow-up and transmit the Dialogue's outcomes to other UN fora, such as the IGF. Convening thematic, multistakeholder, sector-specific consultations with the Co-Facilitators on an ongoing basis is key for nurturing the outcomes and outputs of the Dialogue.

From the perspective of the Global South, the 2030 Agenda remains a foundational reference point, reaffirming state sovereignty over resources and economic activity while committing to implementation consistent with international law and human rights obligations. The Aarhus Convention also offers a valuable model, emphasising public participation, access to information and access to justice - principles that can inform inclusive governance of AI, particularly in environmental and societal domains. The work of the UNESCO Global CSO Network on AI Ethics provides an important parameter for mapping of human rights and global justice considerations across AI value chains and offers a useful reference for the Dialogue. Consideration of the 2030 Agenda ensures that efforts to close gender, racial and socioeconomic divides in AI do not reinforce existing inequalities. In addition and as noted earlier, the Office of the United Nations High Commissioner for Human Rights and UN human rights mechanisms - including Special Procedures and treaty bodies - have developed substantial guidance on applying and enforcing international human rights law in the digital sphere, including AI. The United Nations General Assembly and the broader UN human rights system have also issued specific directions on prohibiting certain harmful AI uses. The Dialogue can help elevate this body of work politically and build upon it. It should also draw on significant contributions from other multilateral actors. In particular, the Council of Europe has advanced one of the few explicitly rights-based AI regulatory approaches, grounded in human rights, democracy and the rule of law. (https://www.coe.int/en/web/portal/-/huderia-new-tool-to-assess-the-impact-of-ai-systems-on-human-rights) A key added value of the Dialogue lies in strengthening alignment across the UN system and broader AI governance processes, promoting greater coherence in line with international law, including human rights obligations.