The UN Environment Assembly on Friday approved its first-ever resolution to address the environmental aspects of Artificial Intelligence (AI), but it did not include a provision to monitor AI systems across their lifecycle. Experts say this approach is essential to understand AI’s water, power and critical minerals consumption.

The resolution proposed by Kenya aims to harness “the opportunities and benefits of artificial intelligence systems in support of the environment and by minimizing its environmental impacts”.

It also requests the UN Environment Programme (UNEP) to produce a report on the “environmental benefits, risks and impacts of artificial intelligence”.

As negotiations progressed over the week in Nairobi, the draft resolution on AI had called for UNEP’s executive director to explore environmental benefits, risks and impacts of artificial intelligence “systems across their lifecycle”.

However, while governments including Kenya, Norway, Colombia and the European Union supported such wording, annotated draft texts showed that Saudi Arabia, Russia and the United Arab Emirates wanted it to be deleted.

When the final resolution was gavelled on Friday, all trace of the AI lifecycle had been removed from the text. References to AI’s water and energy consumption – which featured in previous draft texts – were also removed.

“We cannot talk about sustainable AI without addressing the full lifecycle, from the traceability of critical minerals, to the water used in data centres, to how much renewable energy is being redirected from developing countries to power AI systems in wealthier regions,” said Faith Munyalo, Kenya’s contact point on AI.

Munyalo said that while the adoption of the resolution is an important first step, UNEA must now move forward in future negotiations to address the “blind spots” and deliver stronger language and clearer commitments on lifecycle accountability.

“Sustainability must be built into AI from extraction to disposal, otherwise we risk repeating the same patterns of inequity seen in earlier technological transitions,” she told Climate Home News.

No direct finance expected

As the negotiations reached mid-way point on Wednesday, the AI resolution was on the brink of collapse, essentially over finance, which Saudi Arabia and Iran insisted should primarily flow from developed to developing countries while the UK and the EU argued funding should come from all sources.

Finally, countries landed on a compromise that avoids any obligation for wealthy nations to directly finance AI capacity in the Global South. All countries instead are encouraged to “enhance partnerships” that can mobilise funding, alongside “increased investment, including from the private sector and philanthropy” in AI that supports sustainable development.

AI is finding greater uses in environmental circles, and in developing countries it is already being deployed, boosting funding needs. For example, Sierra Leone in its new NDC climate plan needs almost $7 million, including from donor countries, to build an AI-based climate and weather forecasting system to improve resilience. Also, in Kenya, AI is helping conservationists monitor forest degradation, launch reforestation and predict carbon storage capacity in new forest areas.

Kenya’s Munyalo said most data centres are concentrated in developed countries while Africa lacks the expertise and finance to develop its own AI data systems. A lack of direct funding promises puts the burden back on developing countries and could undermine environmental projects like these, she added.

AI good or bad for energy transition?

Somya Joshi, research director at the Stockholm Environment Institute (SEI), said AI has critical impacts both for climate and biodiversity and needs to be designed in ways that don’t “replicate the same mistakes we made before with extractive technology transitions”.

The debate going forward will need to be informed by science and the environmental impacts along the entire AI value chain, she said, including for water, electricity, critical minerals and rare earths to make semi-conductor chips, as well as pollution and what happens to AI systems at the end of their life.

Joshi said there is a need to prevent growing power demand from AI to reinforce dependency on fossil fuels, which would undermine the clean energy transition.

UN Secretary-General António Guterres earlier this year made a call for Big Tech to power all data centres with 100% renewables by 2030.

Data centres accounted for about 1.5% of the world’s electricity consumption in 2024. But this figure is set to more than double by 2030 as tech giants continue to build out the infrastructure needed to support their power-hungry AI technologies.

While renewable energy sources – combined with batteries – are expected to supply half of the additional electricity, increased demand from data centres will be a “significant” driver of growth for fossil gas and coal-fired generation until the end of this decade, according to the International Energy Agency (IEA).

As the Paris Agreement turns 10, what has it achieved?

Geopolitics limit Nairobi results

The resolution on AI was largely seen by observers as a win for the UNEA, which played out in a tense political environment that limited steps forward on a range of key environmental issues.

The US rejected the outcomes, decrying what it called “climate change theatre”, in line with the denial of climate science by the administration of President Donald Trump and his efforts to thwart climate action.

Behind the scenes, oil-rich Saudi Arabia and Türkiye – host of the COP31 climate talks next year – pushed to water down wording on climate change including the science of melting glaciers.

This rejection of well-established evidence elicited strong criticism from small island nations Fiji and Barbados, as well as the European Union and Australia, in the final session of the conference. Speaking at the closing plenary, the EU delegate said the bloc had arrived at UNEA-7 with high hopes for the environment and multilateralism but have to come to terms with the fact that the Assembly could only achieve good results in some resolutions “and less in others”.

There was also disappointment over a weak resolution on mining and transition minerals, which agreed only on further talks around international co-operation instead of setting up an expert group to identify new instruments to make supply chains greener and more transparent as proposed by Colombia and Oman.

However, fears that some member states would use UNEA as an opportunity to reopen the mandate to negotiate a global treaty on plastic pollution did not come to pass, according to Andrés del Castillo, Senior Attortney at the Center for International Environmental Law (CIEL).

Talks on a new pact were suspended in August as they were unable to reach agreement with fossil fuel-producing countries blocking proposed caps on plastic production – a major market for petrochemicals. They will resume in February with the election of a new chair.

Del Castillo pointed to the ministerial declaration adopted in Nairobi on Friday, which reaffirms countries’ “shared commitment to engaging constructively and actively, with a sense of urgency and solidarity, to conclude the [plastics] negotiations”.

The post UN adopts first-ever resolution on AI and environment, but omits lifecycle appeared first on Climate Home News.

UN adopts first-ever resolution on AI and environment, but omits lifecycle