Lemon OC AB

I speak as a human rights lawyer, and I welcome the establishment of the Global Dialogue as a timely and necessary platform. Today, AI governance is dispersed across national regulations, voluntary standards, and sectoral initiatives. This Dialogue can serve as a unifying space to align these efforts around international law, particularly human rights law, while addressing urgent gaps namely accountability for harm, cross-border impacts, and unequal access to AI technologies.

• Protection and promotion of human rights
• AI capacity-building
• Transparency, accountability, and human oversight
• Safe, secure and trustworthy AI

From my perspective, the Dialogue should prioritize three interconnected gaps. First, the operationalization of human rights and Ethics in AI systems-moving from principles to enforceable safeguards. Second, closing capacity divides, especially for developing countries that risk being rule-takers rather than rule-shapers. Third, ensuring meaningful transparency and accountability across the AI lifecycle, including in private sector systems that have public impact.

Gaps in AI governance are already having tangible impacts across sectors, particularly where rapid technological deployment outpaces regulatory oversight. The most significant challenge is the absence of enforceable, harmonized standards ensuring that AI systems respect fundamental rights such as privacy, non-discrimination, due process, and access to remedy. In practice, this leads to opaque decision-making, biased outcomes, and limited accountability especially in high-risk contexts such as employment, financial services, migration management, and law enforcement. In many jurisdictions, existing legal frameworks are either fragmented or not fully adapted to address the scale and complexity of AI systems. This creates uncertainty for both rights-holders and duty-bearers, and increases the risk that vulnerable groups are disproportionately affected. The lack of transparency in algorithmic systems further undermines the ability to challenge harmful decisions, weakening procedural safeguards that are central to the rule of law. At the same time, there are important opportunities. Emerging regulatory initiatives, coupled with growing awareness among businesses and policymakers, are driving the development of more robust AI governance frameworks. When grounded in international human rights standards, these efforts can enhance accountability, promote ethical innovation, and build public trust. There is also increasing recognition of the need for human rights due diligence in AI development and deployment, aligning technological progress with legal and ethical obligations. Ultimately, the direction of AI governance will determine whether these technologies reinforce existing inequalities or contribute to more equitable and rights-respecting societies.

All stakeholders have a role to play. Member States must lead in establishing legal frameworks and ensuring compliance with international obligations. The private sector must embed human rights due diligence into design and deployment. Civil society and affected communities should be meaningfully included, not only as participants but as co-creators of solutions. Academia and the technical community can provide critical evidence, tools, and standards, while international organizations can facilitate coordination and capacity-building. Finally, the first Dialogue should aim for tangible outputs. These could include a set of guiding principles for human rights-based AI governance, a roadmap for capacity-building partnerships, and a platform for ongoing knowledge-sharing and cooperation. Highlighting existing initiatives, particularly those demonstrating inclusive, rights-respecting AI in practice will be essential to ground the discussion in reality.

From a human rights law perspective, the AI Dialogue should build upon a growing body of international frameworks, multi-stakeholder initiatives, and regulatory instruments that already articulate principles for rights respecting AI. Key among these are the UN Guiding Principles on Business and Human Rights, which establish the corporate responsibility to conduct human rights due diligence; the OECD AI Principles, which promote inclusive, transparent, and accountable AI; and the UNESCO Recommendation on the Ethics of Artificial Intelligence, which provides a comprehensive normative foundation grounded in human rights. At the regional level, the EU Artificial Intelligence Act represents a significant step toward binding obligations, particularly for high risk AI systems, while the Council of Europe Framework Convention on Artificial Intelligence is advancing a legally grounded, human rights based approach beyond the EU. In addition, multistakeholder initiatives such as the Global Partnership on AI and the UNESCO AI Ethics Global Observatory contribute valuable technical expertise, monitoring, and policy coordination. Despite this progress, fragmentation remains a central challenge. Existing initiatives often operate in parallel, with varying levels of enforceability and limited coordination across jurisdictions and sectors. This creates gaps in implementation, oversight, and access to remedy. The added value of the AI Dialogue lies in its potential to bridge these silos by fostering coherence, interoperability, and alignment with international human rights law. It can serve as a platform to translate high level principles into practical, enforceable standards; promote shared accountability between states and companies; and amplify the voices of affected stakeholders, particularly those most at risk of harm. Crucially, it can help ensure that AI governance evolves not only as a technical or economic agenda, but as a matter of legal obligation and human dignity.

To structure discussions effectively, I suggest three thematic clusters. One, "Human Rights and Risk Governance," focusing on due diligence, impact assessments, and remedy mechanisms. Two, "Equity and Access," addressing infrastructure, data governance, and capacity-building. And three, "Interoperability and Standards," ensuring that governance approaches are compatible while respecting diverse legal systems and cultural contexts. These clusters can anchor balanced, solution-oriented discussions. In terms of format, the Dialogue should move beyond general statements and foster practical exchange. This could include moderated roundtables with concrete case studies, scenario-based discussions, and multi-stakeholder working groups tasked with developing specific recommendations. The Scientific Panel's report should not only inform debate but be directly linked to policy labs or breakout sessions that translate its findings into actionable steps. All stakeholders have a role to play. Member States must lead in establishing legal frameworks and ensuring compliance with international obligations. The private sector must embed human rights due diligence into design and deployment. Civil society and affected communities should be meaningfully included, not only as participants but as co-creators of solutions. Academia and the technical community can provide critical evidence, tools, and standards, while international organizations can facilitate coordination and capacity-building.

Global discussions on AI governance continue to underrepresent those most affected by its deployment. This includes communities in the Global South, marginalized racial and ethnic groups, persons with disabilities, workers in precarious or informal sectors, as well as both youth and older persons. Young people are often framed as future stakeholders rather than current rights-holders, despite being among the most exposed to AI-driven systems in education, social media, and digital services. Conversely, older persons are frequently overlooked in AI design and policy processes, leading to exclusion, digital discrimination, and barriers in access to essential services increasingly mediated by automated systems. A core challenge is that participation in AI governance remains concentrated among governments, large technology companies, and technical experts. This creates a structural imbalance where those shaping AI systems are not fully representative of those experiencing their impacts, particularly in contexts involving surveillance, welfare distribution, healthcare, and employment. To address this, inclusion must move beyond consultation toward meaningful participation. This includes investing in capacity-building to enable informed engagement, supporting civil society and community-led organizations, and ensuring accessible, multilingual, and inclusive consultation processes. Mechanisms such as participatory impact assessments, citizen assemblies, and rights-based stakeholder forums can help institutionalize diverse input. Importantly, governance frameworks should recognize affected communities not as passive beneficiaries, but as rights-holders with a legitimate role in shaping the rules that govern AI. Ensuring intergenerational inclusion, bringing both youth and older persons into decision making will be essential to developing AI systems that are equitable, legitimate, and grounded in human dignity.

Meaningful engagement in the AI Dialogue requires formats that go beyond traditional, top down consultations and enable continuous, inclusive, and rights-based participation. One effective approach is the use of deliberative forums, such as citizen assemblies or juries, where diverse stakeholders, including youth, older persons, and marginalized communities can engage with experts, deliberate on real-world AI use cases, and produce informed recommendations. These processes strengthen legitimacy and ensure that governance reflects lived experience. Participatory impact labs are another innovative format, bringing together policymakers, companies, civil society, and affected communities to jointly assess the human rights implications of specific AI systems. By working through concrete scenarios, these labs can bridge the gap between high-level principles and practical implementation. In addition, multi-stakeholder roundtables with equal footing where power imbalances are actively mitigated through facilitation, transparency rules, and clear accountability mechanisms can foster more balanced dialogue between states, industry, and rights-holders. Digital tools can also expand access, including interactive online platforms that allow for global, multilingual input, as well as asynchronous participation to include those unable to attend in real time. However, these must be designed with accessibility and data protection in mind. Finally, youth and intergenerational panels, as well as community-led sessions, can ensure that perspectives often excluded from formal processes are meaningfully integrated. The added value of these formats lies in shifting engagement from symbolic consultation to shared ownership embedding human rights considerations directly into the governance and lifecycle of AI systems.

A key example is the EU Artificial Intelligence Act, which introduces a risk-based approach with binding obligations for high-risk systems, including requirements on transparency, human oversight, and accountability. Complementing this, the UN Guiding Principles on Business and Human Rights provide a widely endorsed framework for integrating human rights due diligence into AI development and deployment, ensuring that companies identify, prevent, and mitigate adverse impacts. At the policy level, the UNESCO Recommendation on the Ethics of Artificial Intelligence sets out comprehensive standards grounded in human rights, including safeguards on data governance, non-discrimination, and environmental sustainability. Similarly, the OECD AI Principles promote transparency, robustness, and accountability, and have influenced national AI strategies across multiple jurisdictions. In practice, algorithmic impact assessments (AIAs) are emerging as an important tool, enabling organizations to evaluate potential human rights risks before deploying AI systems. Some governments and companies are also adopting AI registries and transparency databases, which publicly disclose the use of algorithmic systems, strengthening oversight and public trust. Multi-stakeholder platforms such as the Global Partnership on AI and the UNESCO AI Ethics Global Observatory further support knowledge-sharing, benchmarking, and policy coordination. Together, these approaches demonstrate that effective AI governance is most robust when it combines binding regulation, rights-based due diligence, transparency mechanisms, and inclusive, multi-stakeholder oversight.