Follow-up Mechanism to the Belém do Pará Convention (MESECVI) - Organization of American States (OAS)

A successful first Global Dialogue on AI Governance would be one that moves beyond general principles and produces concrete, actionable outcomes grounded in human rights. First, it should achieve a shared recognition that AI systems are not neutral and can reproduce and amplify existing inequalities, including gender-based discrimination and violence. Integrating a gender perspective, such as the one reflected in the Inter-American Model Law on gender-based digital violence, is essential. This includes commitments to gender-responsive AI governance through impact assessments that consider differentiated risks for women and girls, safeguards against technology-facilitated gender-based violence, and accountability mechanisms for harms. Second, the Dialogue should advance meaningful multi-stakeholder cooperation. Effective AI governance requires the coordinated involvement of States, technology companies, civil society, and technical experts. It should clarify the responsibilities of private sector actors in preventing and mitigating harm, including through safer design choices, transparent content moderation practices, and risk mitigation measures. Finally, success would mean agreeing on practical next steps. This could include the development of shared standards, legislation, voluntary guidelines, and pilot initiatives that can be implemented across regions. It should also promote transparency, access to data for independent research, and stronger international cooperation to address the cross-border nature of AI-related harms. In sum, the Dialogue would be successful if it lays the foundation for inclusive, rights-based, and gender-responsive AI governance that not only fosters innovation but also effectively protects individuals, particularly those most at risk, from emerging harms.

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

Our selection reflects the need to ensure that AI governance frameworks are not only innovative, but also grounded in human rights, effective across jurisdictions, and responsive to real-world harms, including gender-based digital violence, that disproportionately affects women and girls. Safe, secure and trustworthy AI is a priority because AI systems can amplify risks such as online harassment, non-consensual dissemination of intimate content, deepfake abuse, and coordinated attacks that silence women in public life. Ensuring safety and trustworthiness requires integrating risk mitigation measures from the design stage, including safeguards against misuse, content moderation standards, accountability mechanisms, and transparency in how systems operate. Protection and promotion of human rights is central to our work. AI governance must align with international human rights standards, ensuring that technologies do not undermine rights such as privacy, freedom of expression, access to justice, and non-discrimination. A gender perspective, such as that embedded in the Inter-American Model Law on gender-based digital violence, is essential to identify and address how AI can enable, exacerbate, or scale forms of violence against women and girls, and to ensure effective remedies. Interoperability of governance approaches is critical given the cross-border nature of AI systems and digital platforms. Technology-facilitated violence often transcends jurisdictions, making fragmented regulatory responses insufficient. Promoting coherence across frameworks can help close enforcement gaps, strengthen cooperation, and ensure consistent protections. Overall, these priorities reflect a commitment to advancing AI governance that is rights-based, gender-responsive, and capable of preventing, addressing, and remedying violence in digital environments.

Yes. A key cross-cutting and emerging issue that requires greater attention is gender-based digital violence against women, as well as the need for a multidisciplinary approach to effectively address it. Gender-based digital violence against women has become a pervasive and rapidly evolving threat, increasingly amplified by AI systems. These forms of violence not only impact women's safety and dignity, but also restrict their freedom of expression, participation in public life, access to information, and engagement in digital spaces. Despite its scale and severity, this issue is not always sufficiently reflected in AI governance frameworks, which often lack targeted, gender-responsive measures for prevention, mitigation, and accountability. Addressing this challenge requires recognizing that such violence is not merely a byproduct of technology, but a structural issue that intersects with broader patterns of discrimination and inequality. AI systems can exacerbate these harms through biased data, algorithmic amplification, and the misuse of generative tools. Finally, effectively responding to gender-based digital violence against women requires a multidisciplinary approach that integrates legal, technical, policy, and social perspectives. It calls for coordinated action among States, the private sector, civil society, and academia to develop comprehensive, rights-based solutions that address both the technological and societal dimensions of the problem. Incorporating this issue into AI governance discussions is essential to ensure that frameworks are inclusive, effective, and capable of preventing and addressing emerging forms of violence in digital environments.

In Latin America and the Caribbean, governance gaps in safe, secure and trustworthy AI, protection and promotion of human rights, and interoperability of governance approaches are having direct and increasingly severe consequences, particularly in relation to gender-based digital violence against women. There is a growing and urgent demand for regulation across the region. In the absence of clear, enforceable frameworks, AI-enabled technologies are being used to amplify and scale forms of violence such as online harassment, non-consensual dissemination of intimate content, and the creation of synthetic or manipulated images. These harms are not abstract, they have tangible impacts on women's safety, mental health, livelihoods, and participation in public life. In extreme cases, digital violence has escalated to the point of contributing to loss of life. The lack of regulation and oversight also creates significant accountability gaps. Victims often face fragmented or ineffective responses, with limited access to remedies. At the same time, the roles and responsibilities of different actors, particularly technology companies, remain insufficiently defined, allowing harmful content and practices to persist without clear consequences. This challenge is particularly acute in Latin America and the Caribbean, where many countries are small or medium-sized economies. In this context, large technology companies often have greater economic and operational power than individual States, which further complicates efforts to regulate and enforce accountability. Moreover, the absence of interoperable governance approaches across jurisdictions exacerbates these challenges. Digital platforms and AI systems operate transnationally, while regulatory responses remain largely national, leading to inconsistencies, enforcement gaps, and limited cross-border cooperation. These governance gaps underscore the need for comprehensive, rights-based, and gender-responsive regulatory frameworks that clearly define responsibilities, strengthen prevention and mitigation measures, and ensure access to justice. Without such frameworks, the risks associated with AI will continue to expand, disproportionately affecting those already in vulnerable situations.

The AI Dialogue can play a critical role in advancing international cooperation by serving as a platform to bring together multidisciplinary and multi-stakeholder actors to move from principles to concrete steps toward regulation. Effective AI governance cannot be addressed by governments alone. It requires the active participation of technology experts, private sector actors, women's organizations, and broader civil society, alongside international and regional organizations. The Dialogue can help align these actors around shared priorities, foster mutual understanding, and build consensus on rights-based and gender-responsive approaches to AI governance. Importantly, it can also help initiate the urgent process of developing regulatory frameworks that respond to current harms, including gender-based digital violence. By facilitating cooperation across regions and sectors, the Dialogue can support the creation of common standards, promote accountability, and strengthen coordination in addressing the cross-border nature of AI systems. In this regard, the Inter-American Model Law to prevent, punish and eradicate gender-based digital violence against women offers a useful reference for structuring this type of collaboration. It proposes the establishment of multi-actor and interdisciplinary mechanisms that bring together public institutions, the private sector, civil society and technical experts to monitor, guide, and strengthen governance efforts. Such spaces can ensure that diverse perspectives are incorporated, that policies are informed by lived experiences and technical knowledge, and that accountability is shared across actors.

From the perspective of the MESECVI (Follow-up Mechanism to the Belém do Pará Convention) of the Organization of American States, the AI Dialogue should build upon existing regional and international initiatives that have already advanced multi-stakeholder, rights-based approaches to digital governance, particularly those addressing technology-facilitated harms. A key example is the nearly four-year participatory process led by MESECVI to develop the Inter-American Model Law on Gender-Based Digital Violence against Women, adopted in 2025. This process brought together States, International Organisations, civil society, including women's organizations and survivors, academia, and representatives from major technology companies through more than a dozen in-person and virtual consultations across the region. It also included a public consultation that gathered extensive written contributions (100+). This experience demonstrates the feasibility and value of inclusive, interdisciplinary, and multi-actor governance processes in addressing complex digital challenges. While the Model Law focuses specifically on violence against women, the methodology and governance approach it proposes, particularly the creation of multi-stakeholder, inter-institutional mechanisms to oversee implementation and ensure accountability, are broadly applicable to AI governance. It offers a concrete example of how to operationalize cooperation across sectors and levels, integrating legal, technical, and social perspectives. The AI Dialogue can add value by connecting and scaling these types of initiatives at the global level, facilitating cross-regional exchange of good practices, and promoting coherence between emerging frameworks. It can also help bridge existing gaps by translating lessons learned from sector-specific processes, such as those addressing gender-based digital violence, into more comprehensive AI governance strategies. In doing so, the Dialogue can accelerate the development of coordinated, rights-based, and inclusive regulatory approaches that respond to urgent harms while remaining adaptable to the evolving nature of AI technologies.

Different stakeholders can contribute to the AI Dialogue by bringing their specific expertise and perspectives into a structured, multi-stakeholder governance process that moves beyond consultation toward shared responsibility and accountability. From the perspective of the MESECVI, our nearly four-year process to develop the Inter-American Model Law on Gender-Based Digital Violence against Women demonstrated the value of sustained, inclusive engagement among governments, technology companies, civil society, academia, and technical experts. Building on this experience, the AI Dialogue should adopt a format that reflects similar principles. In particular, the Dialogue could draw on the structure proposed in Article 14 of the Model Law, which establishes an inter-institutional and multi-stakeholder mechanism on digital governance. This type of body brings together representatives from executive and judicial branches, political actors, the private sector and internet intermediaries, as well as civil society organizations, including those representing historically marginalized groups, and experts in technology and cybersecurity. Its core function is not only to exchange views, but to monitor implementation, ensure accountability, and integrate human rights, gender, and intersectional perspectives into all actions. Applied to the AI Dialogue, this suggests a structure that includes: • Permanent multi-stakeholder working groups, rather than one-off discussions; • Clear reporting and follow-up mechanisms, including periodic inputs from participating actors; • Spaces for independent contributions, such as civil society reports or expert assessments; • A mandate to generate recommendations and track progress on agreed commitments. This approach would ensure that the Dialogue is not only inclusive, but also action-oriented and accountable, enabling different stakeholders to contribute meaningfully to the development of effective, rights-based, and gender-responsive AI governance.

Women in all their diversity, particularly those affected by gender-based digital violence, remain significantly underrepresented in global discussions on AI governance. Critically, this also includes children and youth, who are among the most exposed to digital environments and emerging AI-related risks, yet are rarely meaningfully included in decision-making spaces. This gap is particularly concerning because children and young people are often early adopters of digital technologies and are disproportionately affected by online harms. At the same time, young women and girls face compounded risks due to gender-based discrimination and violence. Excluding these perspectives results in governance frameworks that fail to fully capture the scope, scale, and lived realities of AI-related harms.

To foster meaningful and dynamic engagement, the AI Dialogue should move beyond traditional panel formats and adopt interactive, multi-stakeholder and action-oriented approaches that enable real exchange, co-creation, and accountability. First, multi-stakeholder roundtables organized around specific themes (e.g., safety, human rights, digital violence) can bring together governments, technology companies, civil society, youth representatives, and technical experts in smaller, solution-focused settings. These spaces should be designed to encourage equal participation and dialogue across sectors. Second, co-creation labs or policy design sprints could be used to develop concrete outputs, such as draft guidelines, risk mitigation frameworks, or accountability tools. These formats allow participants to collaboratively translate principles into practical solutions, combining legal, technical, and social expertise. Third, survivor- and community-informed sessions are essential to ground discussions in lived realities. This could include moderated dialogues with survivors of gender-based digital violence and frontline organizations, ensuring safe and ethical participation, as well as dedicated youth forums to capture the perspectives of children and young people. Fourth, the Dialogue could incorporate structured reporting and feedback mechanisms, such as civil society shadow reports or expert briefings, to complement official inputs and strengthen accountability. Finally, drawing from the experience of the MESECVI and the Inter-American Model Law, the Dialogue could establish ongoing multi-actor working groups that continue beyond the event, tasked with following up on commitments, monitoring progress, and ensuring sustained collaboration. These formats would help ensure that the AI Dialogue is not only inclusive and dynamic, but also capable of producing concrete, rights-based, and gender-responsive outcomes.

From the perspective of the MESECVI (Follow-up Mechanism to the Belém do Pará Convention), one of the most relevant examples of a concrete, rights-based approach to digital governance is the Inter-American Model Law to Prevent, Punish and Eradicate Gender-Based Digital Violence against Women, adopted in 2025. This instrument offers a comprehensive framework that can inform broader AI governance efforts. First, it recognizes that digital technologies, including AI, are not neutral and can facilitate and amplify violence, discrimination, and inequality. As such, it proposes a comprehensive regulatory approach that goes beyond criminalization to include prevention, protection, investigation, attention, sanction, and reparation. Second, the Model Law incorporates specific obligations for digital platforms and intermediaries, including risk mitigation measures, transparency, and responsibilities linked to content moderation and system design. These elements are directly relevant to AI governance, particularly in addressing harms such as deepfakes, algorithmic amplification of abusive content, and the misuse of generative technologies. Third, it advances a multi-stakeholder and interdisciplinary governance model, notably through the creation of inter-institutional mechanisms that bring together public authorities, the private sector, civil society, especially women's organizations, and technical experts. These mechanisms are designed to monitor implementation, ensure accountability, and integrate gender, human rights, and intersectional perspectives. Finally, the Model Law is grounded in a participatory development process, built over nearly four years through consultations with States, International Organisations, civil society, academia, and technology companies. This process itself is a best practice for inclusive governance. Importantly, the Model Law is already moving from norm-setting to implementation. In Latin America, countries such as Uruguay, Ecuador, Brazil, Mexico, and Colombia have begun advancing processes to adapt and incorporate its standards into their national frameworks, demonstrating its practical relevance and regional traction. While focused on gender-based digital violence, this approach offers transferable lessons for AI governance: the need for rights-based frameworks, shared responsibility across actors, and concrete mechanisms for accountability and implementation.