Centre for AI: Social and Digital Innovation, Brunel University of London

We recommend inclusive participation and a detailed roadmap for the Dialogue in July 2026 and beyond, with clear terminological and technological definitions, scope, goals, actionable steps, and measurable indicators for progress between 2026 and 2027. Human-centred, rights-based, and participatory AI governance should be established as a core principle across all stages of AI development, deployment, and evaluation, ensuring that affected communities actively shape governance outcomes rather than being consulted only after implementation. Human-centred AI design should be a clear goal, with best practices embedded from early-stage ideation with technology end users and stakeholders, to design evaluation and iterative refinement. AI should not be reduced to Generative AI, but interpreted broadly to include decision-support systems, homegrown research models for non-commercial research and heritage uses, robotics, immersive technologies, big data infrastructures, and other adjacent technologies. Narrow definitions risk excluding critical governance challenges and limiting meaningful international cooperation. Attribution of responsibilities should be clearly defined to ensure the roadmap is followed and goals are achieved. The UN Independent International Scientific Panel on AI could act as stewards of progress, ensuring evidence is recorded and translated into actionable policy. Methods and approaches for agile governance, as well as implementation and enforcement strategies, should be meaningfully discussed to ensure responsible global adoption and effective enforcement when companies do not adhere to policies. Diversity of participation must be treated as a governance requirement rather than aspirational goal. Special attention should be given to underrepresented stakeholders/perspectives, including Indigenous and marginalised communities and environmental concerns. Discussions should be organised into thematic sub-clusters, grounded in practical case studies, and anchored in human rights. This allows the required depth and diversity of perspectives, so that both risks and opportunities can be effectively discussed. Strategies for Dialogue continuation and roadmap execution should be agreed during the first Global Dialogue event.

• 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

We selected these priorities on the basis that these thematic areas are interdependent and should not be treated as discrete policy domains. These priorities must always be grounded in human rights principles and reinforced within AI system implementation. Across sectors such as healthcare, education, culture and labour markets, AI operates as a complex socio-technical system, where technical performance is inseparable from context, institutional practice, and human interaction. Rapid developments in AI, including decision-support systems, large language models (LLMs), and increasingly autonomous technologies, are reshaping decision-making, service delivery, and organisational processes. However, governance frameworks remain fragmented and opaque. Safety and trustworthiness are not solely technical attributes; they depend on representative data, contextual validity, and equitable performance across diverse populations and use environments. Social, cultural, and linguistic factors shape how AI systems function in practice, particularly in cross-border and multilingual contexts. Meanwhile, transparency and accountability are constrained by limited visibility into system behaviour and unclear allocation of responsibility across developers, deployers, and institutions. The technical development of AI systems progresses significantly faster than the ability to understand AI behaviour and social impacts. Society must be educated and equipped with mitigation strategies to enhance AI system behaviour continuously. Classical human oversight remains under-specified and risks becoming ineffective without appropriate system design, interpretability, and user capability. Economic asymmetries and uneven global participation raise concerns regarding fairness, inclusion, and equitable benefit-sharing. Meaningful stakeholder participation in policy- and decision-making is mandated by human rights frameworks such as cultural rights. Public trust is also achieved when affected communities are involved in co-designing, deploying, and evaluating AI systems rather than consulted only after implementation. We therefore advocate for an integrated interdisciplinary co-design governance approach that advances these priorities collectively, ensuring AI systems are technically robust, socially legitimate, contextually appropriate, accountable, human-centred, and fully aligned with human rights.

Several cross-cutting and emerging issues warrant further attention beyond the themes identified, particularly in relation to the systemic governance of AI across its lifecycle, infrastructure, and environmental footprint. First, there is a need to strengthen lifecycle governance frameworks, including mechanisms for post-deployment monitoring, updating, auditing, and withdrawal of AI systems. Closely related is the persistent gap in context-sensitive validation, as systems are frequently developed in one environment and deployed in others without sufficient adaptation or evaluation. This is compounded by the absence of harmonised evidence standards in high-impact domains, as well as limited clarity on accountability once systems are operational. Second, emerging epistemic risks associated with large language models require greater attention. These systems may generate outputs that are coherent and authoritative but factually incorrect or unethical, introducing subtle risks that are difficult to detect, particularly in time-sensitive or high-stakes environments. Finally, environmental sustainability must be treated as a core cross-cutting dimension, encompassing both the environmental costs of AI systems from training to inference ("Green AI")- including energy use, water consumption, unsustainable mining of critical minerals, and e-waste- and their potential for AI to research and support climate and sustainability objectives ("AI for Green"). These issues highlight the need for integrated governance and research approaches that address AI as a lifecycle socio-technical system with environmental, institutional, and infrastructural dimensions.

Governance gaps and rapid advances in AI are having increasingly visible socio-economic, cultural, and environmental impacts globally, particularly in creative industries and local communities. The rapid expansion of data centre infrastructure, while enabling AI development, is placing pressure on local environments through energy demand, water usage, land allocation and mining. These environmental externalities are often locally concentrated, raising concerns about sustainability and community impact, while remaining insufficiently addressed within existing governance frameworks. Significant disruption of future of workforces is also evident across industries, particularly creative and digital sectors. Across the creative industries, widespread adoption of generative AI has contributed to workforce displacement, with reported redundancies as automation is introduced into design, art, and narrative development workflows. However, some studios are partially reversing these decisions after AI deployment proved detrimental to production quality, leading in some cases to rehiring and workflow restructuring. Furthermore, in the music sector, AI-generated content integrated into streaming platforms has raised concerns regarding attribution, value extraction, consent, remuneration and cultural displacement. The use of AI models trained on existing copyright works risks reinforcing deplatforming dynamics, where human creators face reduced visibility and economic return. While regulatory frameworks such as the EU AI Act and the UK's principles-based approach are beginning to address these issues, implementation and enforcement strategies remains uneven, particularly cross-border. Many organisations continue to lack mature governance structures, resulting in fragmented oversight of AI deployment, monitoring, and accountability across sectors. These concerns were also reflected in the BRAID researchers' submission to the UN Expert Mechanism on the Right to Development, which highlighted the intersection of AI governance gaps with environmental impacts on local communities and the uneven distribution of benefits and harms across the digital economy. These points underscore both significant risks and emerging opportunities for more responsible, context-sensitive AI governance.

The AI Dialogue can play a central role in advancing international cooperation on AI governance by functioning as a global convergence platform that connects fragmented initiatives and oversight bodies, and translates thematic discussions into coordinated action. A key contribution would be to strengthen alignment across existing international repositories and governance efforts, including UNESCO's AI knowledge resources and related UN, OECD, and regional initiatives. Rather than duplicating existing work, the Dialogue can improve interoperability between these systems, enabling more coherent knowledge exchange and supporting shared reference points for policymaking. It can also facilitate the development of shared approaches in emerging cross-cutting domains, particularly where global standards remain underdeveloped. This includes environmental measurement of AI systems, where consistent methodologies are needed to support comparability and accountability across jurisdictions and sectors. In addition, the Dialogue can provide a structured space for addressing unresolved transnational governance questions, including those linked to data governance, open AI ecosystems, intellectual property, and cultural production. This is especially relevant in the context of ongoing international discussions within the World Intellectual Property Organisation (WIPO), whose work on exceptions and limitations for cultural heritage and the AI Conversation should connect to the Dialogue. A further important role is the expansion of meaningful global participation in AI governance. This includes improving the inclusion of stakeholders who are currently underrepresented in international debates, particularly from the Global South, civil society, and affected communities, thereby strengthening the legitimacy and balance of global decision-making. This is particularly important where unequal access to compute infrastructure, regulatory capacity, and technical resources limits meaningful participation in global AI governance. Therefore, the AI Dialogue can serve as an enabling mechanism and platform that connects fragmented governance efforts, supports the development of shared international reference frameworks, and strengthens inclusive participation in shaping the future of AI governance.

The AI Dialogue should build upon and connect existing international, regional, and research-based initiatives in order to strengthen coherence in global AI governance while avoiding work duplication. At multilateral levels, initiatives include UNESCO's Global AI Ethics and Governance Observatory, (e.g. AI, culture and intellectual property sub-group and its proposed repository), and UN Special Rapporteur on cultural rights work on AI and creativity. Connections with OECD AI Principles and Global Partnership on AI (GPAI) would strengthen policy alignment, shared standards, and practical governance coordination across jurisdictions. Engagement with the UN Special Rapporteur in the field of cultural rights and the UN Expert Mechanism on the Right to Development would be valuable in ensuring that the Dialogue builds on existing work and that AI governance is framed within broader questions of human rights, equity, inclusion, and sustainable development. At the research and policy interface, programmes such as UKRI ESRC and AHRC-funded programmes including BRAID, TaNC, DiSSCo, RICHeS and the Digital Good Network, provide important empirical and interdisciplinary insights into the societal, cultural, and economic implications of AI, particularly in relation to data, governance, and digital infrastructures. The Dialogue should engage with emerging regional innovation ecosystems such as the European Union's AI Factories and AI Antennas under the 2024 Innovation Package. These infrastructures are designed to support SMEs, startups, and researchers in developing and deploying AI systems, and represent a significant investment in AI capacity-building, innovation, responsible deployment across Europe. The added value of the AI Dialogue lies in its ability to connect these diverse initiatives into a more coherent global governance ecosystem, facilitating structured exchange between normative frameworks, research evidence, and technical infrastructures. It can serve as a bridging platform between policy, innovation, and rights-based approaches, enabling mutual learning across regions while strengthening inclusivity and interoperability in AI governance.

Attribution of responsibilities should be clearly defined to track progress and ensure the Dialogue goals are being met. The UN Independent International Scientific Panel on AI could function as stewards of goal achievement for the Dialogue, ensuring the surfacing evidence is recorded and translated into actionable policy. Its role should include monitoring progress, identifying implementation gaps, and ensuring that commitments are translated into measurable policy outcomes rather than remaining at the level of consultation alone. Diversity of participation should be a priority, with special attention to underrepresented stakeholders and perspectives, including Indigenous and marginalised communities and environmental concerns. Discussions should be organised into thematic sub-clusters and grounded on practical case studies, allowing greater depth, evidence-based discussion, and clearer identification of both risks and opportunities across sectors. Stakeholder participation should include academia, civil society, government, private sector, public sector such as healthcare representatives, social, technical and scientific communities, heritage and cultural organisations, frontline professionals, and local community groups, particularly those traditionally underrepresented in global governance processes. The first Global Dialogue event in July 2026 is only a starting point, and the UN should engage in a continuous dialogue with participatory design with stakeholders, and regional expansion, including Global South countries, to fully engage local communities and regional perspectives. Hybrid and multilingual participation formats, including regional town halls and asynchronous digital engagement, should be used to ensure accessibility across different geographies, languages, and resource settings, to ensure inclusive and meaningful global participation. Work should be done to ensure that stakeholders are made fully aware of and educated on the risks and opportunities offered by AI so that their Dialogue/townhalls engagement is informed. Grassroots engagement should be actively supported through partnerships with local civil society organisations and community actors, ensuring participation is meaningful and capable of influencing final outcomes.

Several voices remain significantly underrepresented in global AI governance discussions, particularly those most directly affected by cultural, environmental, labour, and rights implications. Meaningful inclusion could be strengthened through structured participatory mechanisms, compensated engagement models, sector-specific consultations, and proactive integration of underrepresented communities into AI governance processes. Underrepresented voices are stated as follow: Indigenous and Global South communities: despite being disproportionately affected by extractive data practices, infrastructure expansion, and unequal access to AI governance, compute resources, and regulatory capacity. Their inclusion is essential for equitable and globally legitimate governance. Communities located near AI infrastructure, particularly data centres, are also underrepresented despite experiencing significant environmental impacts, including increased energy and water consumption, pollution, and rising local resource pressures. Individual artists and creators: despite growing evidence that their work is used to train generative AI systems without consent or fair compensation. This spans music, visual arts, and performance, where the replication of voices, styles, and digital likenesses raises complex questions of authorship, ownership, consent, dignity, and posthumous representation. Recent developments in film, music, and gaming sectors illustrate emerging concerns regarding synthetic personas and the erosion of safeguards for living creative labour. Authors of literary works, such as novelists, and academic writers, are also underrepresented. Moreover, intellectual property stakeholders beyond traditional copyright holder, such as those concerned with trademark protection and brand integrity are increasingly affected by generative AI systems. Vulnerable communities, such as children and elderly people, and healthcare workers and patients are more susceptible to harm and are directly affected by AI-enabled decision-making and service delivery. Data centre workers (including AI moderation and data labelling): although they have an essential role in system development, they are frequently outsourced across global supply chains, exposed to harmful content under high productivity pressures, and often lack adequate occupational or psychological support.

Innovative engagement formats for the AI Dialogue should prioritise interactive, practice-based, and inclusive approaches that move beyond traditional consultation models and enable more meaningful deliberation on complex governance challenges. Engagement must acknowledge systemic barriers and ensure that underrepresented groups are enabled through inclusive practices. Case-based discussions grounded in real-world AI deployments- such as clinical decision-support systems, wearable technologies, large language model applications in service delivery, and AI deployments in heritage and research - can help anchor discussions in practical contexts. This allows participants to directly engage with governance dilemmas as they arise in operational environments, including issues of safety, accountability, and transparency. Multi-stakeholder scenario workshops can further enhance engagement by bringing together policymakers, technical developers, domain experts, and affected stakeholders, such as clinicians, patients, creators, researchers, heritage practitioners and community representatives. These formats enable structured exploration of trade-offs and unintended consequences, particularly in relation to ethics, responsibility, oversight, and safe implementation. To ensure that lived experience is meaningfully integrated, structured input from frontline settings should be incorporated, including contributions from healthcare and heritage professionals, community-based providers, and other end-users of AI systems. This helps ensure that governance discussions reflect operational realities rather than abstract assumptions. Hybrid and asynchronous participation models are also essential to broaden accessibility and global inclusion. Multi-modal multilingual digital platforms should be used to enable sustained engagement beyond live sessions, allowing participation across time zones and resource-constrained settings. Iterative engagement mechanisms with clear feedback loops should be embedded into the Dialogue structure. This ensure that contributions are not only collected but visibly reflected in evolving outputs, strengthening transparency, continuity, and trust in the process.

UNESCO's subgroup on AI, culture and intellectual property, including its emerging repository and related consultation and tracking mechanisms, provides an important initiative for knowledge consolidation, participatory governance, and evidence-based international coordination. In AI's fast-paced and ever-changing governance landscape, relying on static documentation creates visibility gaps, whereas "living" trackers of cases, regulation and governance act as real-time resources that evolve with the work, ensuring decisions are based on current reality rather than outdated snapshots. These efforts support transparency, comparative learning, and the development of shared reference resources across jurisdictions. International soft-law and multi-stakeholder frameworks such as the OECD AI Principles and the Global Partnership on AI (GPAI) further contribute to global coordination by promoting shared values, policy alignment, and practical governance tools that can be adapted across different regulatory environments. The European Union's AI Act represents a significant development in risk-based regulation, introducing differentiated obligations based on system risk, including requirements for transparency, human oversight, and post-deployment monitoring for high-risk applications such as clinical AI. This is complemented by the UK's principles-based regulatory approach, which emphasises flexibility, innovation, and context-sensitive oversight rather than prescriptive rulemaking. EU's 2024 AI Innovation Package introduces AI Factories and AI Antennas to strengthen AI development capacity across Member States. These infrastructures are designed to support SMEs, startups, and researchers in developing, training, and deploying AI systems, thereby linking governance with innovation ecosystems and competitiveness objectives. In the health domain, regulatory guidance from the Medicines and Healthcare products Regulatory Agency (MHRA) further reinforces the importance of lifecycle governance, particularly for AI as a medical device, highlighting the need for continuous monitoring and post-market oversight. Collectively, these initiatives demonstrate that effective AI governance requires risk-based regulation, continuous lifecycle oversight, multi-stakeholder coordination, and infrastructure development as complementary components of effective AI governance.