Although Artificial Intelligence (AI) had been considered in previous COPs1 , COP30, which took place in Belém, Brazil, in November 2025, marked a significant new phase in climate discussions. For the first time, AI was systematically included in the COP’s Action Agenda as a strategic theme. This incorporation was promoted through official events, thematic days, institutional initiatives, and partnerships. There was considerable enthusiasm for AI's potential to help address the climate crisis. Additionally, the conference formally announced the establishment of the AI Climate Institute (AICI), highlighting the growing role of AI in climate governance.
However, despite the strong enthusiasm around AI's promises to help tackle climate change at COP30, very little attention was given to the other side of the AI ecosystem: its environmental impacts. Only a few side events and press conferences drew attention to how AI models and the infrastructures that power them are responsible for emitting high levels of CO₂ into the atmosphere and have also led to a heightened demand for minerals, water, and energy.
This absence is symptomatic of a broader institutional blind spot. According to recent analyses, the digital sector — including AI infrastructure, data centers, and Big Tech operations — already accounts for between 1.5% and 4% of global greenhouse gas emissions. The surge in electricity demand from data centres worldwide is set to more than double by 2030 to around 945 terawatt-hours (TWh), slightly more than Japan's total electricity consumption today, according to the International Energy Agency . Despite their rapidly growing contribution to global energy demand, water use, and material extraction, these impacts remain unreported in the vast majority of NDCs3. This lack of transparency and accounting risks turning AI into a new invisible driver of climate disruption.
In an effort to draw attention to these socio-environmental impacts, activists turned to one of the few available spaces in the negotiation zone, using the UNFCCC press conference “AI’s Global Threats to Climate & Environmental Justice” to expose how these impacts are still being ignored in multilateral negotiations. The issue was also addressed in activities at the People’s Summit.
At the end of COP30 and in the context of the political discussions that must continue to be developed in future editions of the COP, we would like to express the following concerns regarding the public discourse on AI in the context of the climate and ecological crisis:
1. Artificial Intelligence is not a techno-solution to the climate and ecological crisis; moreover, AI increases the use of fossil fuels, raises greenhouse gas emissions, and thus jeopardizes the climate goals of countries with the highest concentration of AI data centers, such as China, the United States and the European Union. COP climate policies cannot be based on marketing discourse, lobbying, or magical thinking promoted by technology companies, but rather on current independent scientific evidence
2. Artificial Intelligence is not just another natural resource or an inevitable force. Its use, adoption, and marketing in all aspects of political, social, and economic life is driven by its owners, a handful of large and powerful technology companies (concentrated mainly in two countries, the United States and China) whose incentive is to expand their capital, not to mitigate the climate and ecological crisis. COP climate policies cannot be designed to serve the economic well-being of this handful of already powerful companies: this encourages the concentration of power and dangerously strengthens their role, especially in other low-income and developing countries.
3. AI generates socio-environmental impacts far beyond CO₂ emissions. As multiple international reports based on scientific evidence show, AI is an industry that requires numerous minerals, large amounts of land, and vast quantities of fresh water and energy, which is causing a series of socio-environmental impacts around the world that go beyond Scope 1 CO₂ emissions - also demanding a serious accounting of Scope 3, the category that exposes the full lifecycle impacts across mining, supply chains, manufacturing and end-of-life. Yet COP30’s outcomes did not meaningfully incorporate these impacts, leaving a major gap in how countries assess and report the climate footprint of digital infrastructure. Looking forward, it is essential that national climate commitments (Nationally Determined Contributions - NDCs) explicitly include the emissions and resource use associated with data centers and AI supply chains, ensuring transparency and accountability in a sector whose climate impact is rapidly expanding. We are concerned that decision-makers believe that these impacts can be miraculously solved by technological innovation alone, which the evidence rules out, for example, given Jevons' paradox in AI.
4. AI’s hunger for energy threatens a just energy transition. As one of the most energyintensive industries of the 21st century, the genuine interest of the companies behind AI at the COP is to secure access to fossil fuels in the short term and renewables in the medium term, the latter being considered a techno-solution to their CO₂ emissions, ignoring the social, economic, and environmental costs that renewable energy production currently entails, especially in communities that have not caused the climate and ecological crisis. The AI's appetite for renewable energy is such that, without political and democratic mediation, we denounce that the energy transition, especially in developing countries, will be designed for the needs of a handful of foreign technology corporations rather than for local communities and industries.
5. Governments must protect their people and ecosystems, not the industry interests. We urge decision-makers in national governments, particularly in developing countries participating in the COP, to reaffirm their commitment to scientific evidence and the well-being of their communities, biodiversity, and local industries. It is essential not to adopt AI uncritically. We are at a critical juncture in addressing the climate and ecological crisis, and any enhancement of AI without proper regulatory, socioenvironmental and ethical checks will only strengthen the power of global tech corporations, ultimately undermining climate ambitions worldwide.
Signed by:
• The Latin American Institute for Terraforming
• Law and Technology Research Institute of Recife (IP.rec)
• Coding Rights
• Laboratory of Public Policy and Internet – LAPIN
• Brazilian Institute for Consumer Protection – Idec
• Heinrich Böll Foundation
