United Nations Office for Outer Space Affairs

A successful first Global Dialogue on AI Governance would, in our view, be one that moves from broad principles to practical, inclusive, and implementable outcomes. For UNOOSA, success would mean that the Dialogue clearly recognizes that AI governance must also address the space and Earth-observation domain, where applications are increasingly high-stakes, transboundary, and closely linked to disaster management, climate action, and sustainable development. First, the Dialogue should produce a shared understanding that trustworthy AI requires risk-based safeguards, including transparency, traceability, meaningful human oversight, and fail-safe mechanisms for critical applications. This is especially important for AI used in space operations and AI-derived geospatial products that can influence public decisions. Second, success would mean concrete follow-up pathways: common benchmarks, practical guidance, and multistakeholder cooperation on issues such as evaluation of geospatial foundation models, data provenance, content integrity, and responsible data governance. Third, the Dialogue should ensure that developing countries are not left behind. A meaningful outcome would include stronger commitments on capacity-building, equitable access to AI-ready Earth observation data, computing resources, open models where appropriate, and knowledge-sharing that enables all countries to benefit. Finally, success would mean establishing a durable bridge between the UN's emerging AI governance architecture and existing multilateral space governance processes, including COPUOS. The UN's roadmap shows that this first Dialogue is intended as the opening step in a longer process, with the Global Dialogue in Geneva on 6–7 July 2026 following consultations and a formal programme published in May 2026. For UNOOSA, success is therefore not only a strong event, but a credible foundation for sustained, responsible, and inclusive governance of AI in space for the benefit of all humankind.

Safe, secure and trustworthy AI;Social, economic, ethical, cultural, linguistic and technical implications of AI;Transparency, accountability, and human oversight;Open-source software, open data and open AI models;

UNOOSA selected these four priorities because they correspond most directly to the urgent governance gaps identified in our policy brief on responsible AI in space and Earth observation. The brief stresses that AI in this domain is increasingly used in high-stakes, time-critical contexts, from on-board autonomy to geospatial analysis for disaster response, where failures, opacity, bias or misuse can have real operational and societal consequences. Safe, secure and trustworthy AI is a priority because space operations require explainability, traceability, fail-safe behaviour and risk-based safeguards, especially where AI can affect spacecraft state, mission continuity or public decision-making. Transparency, accountability, and human oversight are equally essential, since UNOOSA's brief calls for context-appropriate human control, auditability, safety cases, incident reporting and clear governance for dual-use implications. Social, economic, ethical, cultural, linguistic and technical implications of AI is also central to UNOOSA because uneven access to data, compute, skills and representation can deepen global inequities. Our brief highlights the need for fairness, inclusivity and capacity-building so that developing countries and emerging space actors can participate meaningfully and benefit equitably from AI-enabled space applications. Finally, open-source software, open data and open AI models is a priority because openness, where appropriate and accompanied by safeguards, can widen access to AI-ready Earth observation data, benchmarks and models, support scientific collaboration, and reduce capability gaps across regions. At the same time, UNOOSA emphasizes that openness must be paired with responsible licensing, provenance, and protection against manipulation or misuse.

Yes. In UNOOSA's view, several cross-cutting and emerging issues deserve more explicit attention. First, AI governance for space and Earth observation should address the distinct characteristics of geospatial foundation models. These models are advancing faster than they are being independently evaluated, and governance should therefore include not only accuracy, but also transferability across regions and sensors, interpretability, robustness, social impacts, and environmental footprint. Second, data provenance, authenticity and protection against manipulation should be treated as a standalone priority. In the geospatial domain, AI can make it easier to alter, mislabel or misinterpret imagery and derived products. This creates risks for misinformation, operational error and loss of trust, especially when AI outputs inform disaster response, climate services or other public decision-making. End-to-end provenance, watermarking, cryptographic verification and trusted data-sharing arrangements are therefore increasingly important. Third, dual-use implications merit stronger emphasis. Some AI capabilities developed for civilian space and Earth-observation applications may also have security-sensitive implications. Governance frameworks should therefore encourage transparency and safeguards that protect civilian and public-interest uses while respecting existing international norms for the peaceful use of outer space. Fourth, the environmental sustainability of AI-enabled space systems should be more visible as a cross-cutting issue. UNOOSA's brief highlights the need to assess energy use and environmental impacts across the full EO+AI lifecycle, from model training and computing infrastructure to satellite manufacturing, launch and end-of-life management. Finally, equitable access to data, compute, benchmarks and skills should be recognized as a systemic issue across all themes. Without this, AI governance risks reinforcing existing inequalities and leaving many developing countries behind.

For UNOOSA and the wider space and Earth-observation sector, the main challenge is that AI capabilities are advancing faster than governance, evaluation, and access arrangements. In practice, this affects high-stakes uses across the value chain, from on-board autonomy and mission operations to geospatial analysis, dissemination, and decision support. When systems are insufficiently transparent or weakly tested, this can create operational risk, reduce trust, and complicate accountability in time-critical contexts such as disaster response. A second major challenge is uneven access. The policy brief notes that many countries still lack equitable access to AI-ready Earth observation data, computing resources, benchmarks, and skilled personnel. If left unaddressed, these gaps can deepen existing inequalities, especially for developing countries and emerging space actors, and limit their ability to benefit from AI-enabled public-good applications. A third challenge concerns data integrity and openness. Open data, open models, and reproducible workflows create major opportunities, but they also require safeguards. In the geospatial domain, AI can make it easier to modify, mislabel, or misinterpret products, which raises risks of misuse, misinformation, and loss of confidence unless provenance, licensing, traceability, and audit trails are strengthened. At the same time, the opportunities are significant. AI can improve the speed and usefulness of Earth-observation applications for disaster risk reduction, climate services, agriculture, and related decision-making. UNOOSA's own service offer already reflects this demand through expertise-on-demand for AI-driven EO analytics and governance support for sensitive use cases. For UNOOSA, this creates a clear opportunity to help Member States translate principles into practice through capacity-building, practical guidance, open and responsible data-sharing, and support for common standards that keep AI in space safe, transparent, inclusive, and aligned with the peaceful uses of outer space. One of the key challenges in the space domain is determining the appropriate intergovernmental forum to address the interlinkages between space and artificial intelligence. Traditionally, the Committee on the Peaceful Uses of Outer Space (COPUOS) would be the natural venue for such discussions. However, given the rapidly evolving space environment, the committee's agenda is already under considerable pressure from a wide range of competing priorities. A further constraint is that the current pool of experts and Member State representatives within COPUOS may not yet possess the specialised background required to engage substantively with the intersection of space and AI. The central question for the future is therefore whether space-AI issues should be addressed within COPUOS itself, or whether space considerations should instead be integrated into broader AI-focused dialogues elsewhere in the multilateral system — with UNOOSA serving as the dedicated representative and advocate for the space dimension within those processes

UNOOSA started the work under Ensuring Responsible AI in Space and Earth Observeration and released the following Policy brief: https://www.unoosa.org/res/oosadoc/data/documents/2025/p/-_0_html/UNOOSA-Ensuring-Responsible-AI-in-Space-and-Earth-Observation.pdf