AI in or for environmental governance? Ideas, sustainability pathways and geopolitics
Marie Francisco
The recent progress of large language models, and generative artificial intelligence (AI) more generally, has been the object of many discussions that have inspired both awe and fear. One point seems to rapidly gather consensus: AI must be regulated.
My PhD project (Francisco 2025) aimed to analyse the international governance of AI from a sustainability and geopolitical perspective. As a matter of fact, pressing challenges spanning algorithmic bias, cybersecurity, impacts on the labour market, democracies and data ownership are important and necessary aspects of making sure AI does no harm. But in addition, questions about the water and energy consumption of generative AI and the construction of data centres show one thing: the successful governance of AI cannot miss out on ensuring AI is sustainable and contributes to sustainability when used.
In addition, how AI governance is understood will impact who benefits, how, and where. This is why political ideas and the power they hold must be interrogated, and this what I have shed light on with my research.
AI governance: promoting AI for sustainability and for the benefit of all
Studying international and national policies about AI and digitalisation highlights three main tensions in AI governance (Francisco & Heintz, 2026). The first one is how to balance the need for global regulations, which may require flexibility, with developing policies that are precise enough to have an impact when implemented. The second tension lies between long-term and short-term benefits and risks, as governing AI addresses different temporality. The final tension finds itself between the interests that countries, and polities like the European Union (EU), may have in ensuring technological sovereignty and advantages, and the need for international collaboration to develop successful AI models. In this context, international organisations like the United Nations (UN), the OECD or the African Union have attempted to play a role as platform for collaboration (Francisco & Heintz, 2026).
Against this background, policy documents from the UN, the EU or the World Economic Forum show two approaches to the governance of AI in a sustainability context (Francisco & Linnér, 2023). The first approach aims to encourage the use of AI for sustainability purposes. We identified that the use of AI is expected to contribute to better data collection and analysis, with examples such as diseases outbreak, climate change patterns, or energy systems. The idea underlying this argument is that sustainability challenges can be solved with high quality and rapid knowledge of what is happening, but also more accurate future predictions. In addition, AI is expected to enhance automation. Automation, in combination with more evidence-based decision-making, is also expected to bring about greater efficiency and reduced inequalities, especially for women and girls, and populations in developing countries (ibid.). But on the other hand, many of the policies from international organisations and states I analysed call for a global common framework with room for flexibility, to be complemented by more precise, and national regulations (Francisco & Heintz, 2026; Francisco & Linnér, 2023). The EU, the UN, and the OECD in particular have been calling for trustworthy, ethic-based or human-centred approaches in AI development (ibid.).
The hidden geopolitics of AI governance
The ideas behind international and AI policies orient the debate and goals of AI governance. The AI governance policies I analysed reinforce the focus on data-based and technologically-enabled environmental action, which is typical of the ecomodernist and green governmentality discourses (Francisco, 2025; Francisco & Linnér, 2023). In this respect, the reliance on technical solutions for environmental governance are in fact common discourses in environmental governance, which seem at odds with the otherwise expectation that new technologies are transformative (Ulnicane et al., 2022).
Moreover, the current AI governance landscape tends to sideline the interest of the Global South (Francisco, 2025; Francisco & Heintz, 2026). Of course, there are calls for the inclusion of developing countries and marginalised groups, especially by the UN, the OECD, and the African Union (Francisco & Heintz, 2026; Francisco & Linnér, 2023). But when looking more in detail at what the policies focus on, structural inequalities and the global political economy of AI, for example the extraction of critical minerals for building computer hardware, or the ethical challenges linked to data annotation and extraction, are seldom points of discussions. It means that while international policies recognise the need to actively include marginalised populations, many of their specific needs go unacknowledged. In addition, the AI policies I analysed implicitly create power imbalances, with a centre composed of the United States, China, and leading organisations like the UN, the OECD and the EU, while other states are relegated to more peripheral roles (Francisco, 2025).
AI governance and climate security: different perspectives
The complex geopolitical and sustainability challenges surrounding AI lead to the question of climate security. Something does not simply happen to be a security problem; it is usually the result of political debates and efforts by political actors to turn a topic into a matter of security. With global environmental change, the climate has become an important element of security, with different understanding of what should be secured and how (McDonald, 2013).
Depending on what is meant by climate security, the understanding of how AI can be included is likely to differ in four main ways (Francisco, 2023). First, AI could be included in efforts to foster national security. That is, the political focus would be on the use of technologies such as drones and computer vision in climate-induced conflicts (or conflicts perceived as such). A national understanding of security would also focus on the use of online propaganda and misinformation, as opinions on climate change are tied to strong political identities. Second, an international understanding of climate security would shed light on how AI could be used by international organisations in their work to foster sustainable development, but also on the use of AI by private companies. Third, a human understanding of climate security would support efforts to use AI-technologies for sustainable development and the betterment of humans more generally, on the condition that risks are also addressed. Finally, an ecological view on climate security would allow for an exploration of how AI and digital technologies affect the way we see and relate to the world, which could open avenues on the sustainable governance of AI (Francisco, 2023, 2025).
Dr. Marie Francisco is a postdoctoral researcher at the Department of Thematic Studies- Environmental Change of Linköping University, Sweden. With a background in International Relations and Social Environmental Science, her research focuses on the environmental policy discourses about artificial intelligence and digital technologies at the international and national level.
References
Francisco, M. (2023). Artificial intelligence for environmental security: National, international, human and ecological perspectives. Current Opinion in Environmental Sustainability, 61, 101250. https://doi.org/10.1016/j.cosust.2022.101250
Francisco, M. (2025). Making Sense of Artificial Intelligence for Global Environmental Governance: Ideas, Power, and Policy Pathways. Linköping University Electronic Press. https://urn.kb.se/resolve?urn=urn:nbn:se:liu:diva-213333
Francisco, M., & Heintz, F. (2026). 14- The geopolitics of AI in global environmental governance. In B.-O. Linnér, T. Bennich, & H. Carlsen, Handbook on the Geopolitics of Sustainability (pp. 167–178). Edward Elgar Publishing. https://doi.org/10.4337/9781035342549.00026
Francisco, M., & Linnér, B.-O. (2023). AI and the governance of sustainable development. An idea analysis of the European Union, the United Nations, and the World Economic Forum. Environmental Science & Policy, 150, 103590. https://doi.org/10.1016/j.envsci.2023.103590
McDonald, M. (2013). Discourses of climate security. Political Geography, 33, 42–51. https://doi.org/10.1016/j.polgeo.2013.01.002
Ulnicane, I., Knight, W., Leach, T., Stahl, B. C., & Wanjiku, W.-G. (2022). Governance of Artificial Intelligence: Emerging International Trends and Policy Frames. In The Global Politics of Artificial Intelligence. Chapman and Hall/CRC. https://doi.org/10.1201/9780429446726-2