UN Launches AI Environmental Transparency Initiative, Pressing Companies to Disclose Their Footprint

UN Launches AI Environmental Transparency Initiative, Pressing Companies to Disclose Their Footprint

The United Nations is no longer asking artificial intelligence companies to talk about their carbon footprint. It is asking them to confess it.

In a special address at London Climate Action Week on June 23, UN Secretary-General Antonio Guterres launched the AI Environmental Transparency Initiative, a direct appeal to the world's largest AI firms to measure and publicly disclose the full environmental cost of their systems, including the carbon they emit, the water they consume, and the land they occupy. He coupled the demand with a clean-energy target: power every data centre with renewable electricity by 2030.

"No more hidden costs. No more shifting the burden onto those least able to bear it. It is time to come clean," Guterres said. "If AI is to help build a better future, it must be honest about what it costs us now."

What the initiative asks for

The initiative is, at its core, a disclosure regime built on persuasion rather than law. It calls on major AI companies to publish the complete environmental footprint of their systems across three dimensions that have historically been reported, if at all, in fragments: carbon emissions, water withdrawals, and land use. It also asks those companies to commit to sourcing all data-centre electricity from renewables by the end of the decade.

Guterres framed the secrecy itself as the harm. "Despite these obvious concerns, communities are often left in the dark about the environmental impact of the infrastructure rising around them," he said, pointing to the towns and regions where data centres are being built faster than local residents can learn what they will consume.

The pitch is notable for what it does not contain: enforcement. There is no penalty for non-disclosure, no standardized reporting template mandated by treaty, and no audit mechanism. It is a transparency norm the UN hopes the industry will adopt, and that governments and the public will then use as leverage.

The numbers behind the alarm

The initiative draws directly on a report released earlier in June by the UN University Institute for Water, Environment and Health (UNU-INWEH), titled "Environmental Cost of AI's Energy Use: Carbon, Water and Land Footprints." The study, led by UNU-INWEH Director Kaveh Madani, attempts to make AI's physical demands legible.

Its projections are stark. Global data-centre electricity use could reach roughly 935 terawatt-hours by 2030, about 3 percent of projected world electricity, producing nearly 399 million tonnes of CO2. AI's share of that data-centre energy, the report estimates, will climb from around 20 percent today toward 40 percent by 2030. The associated water footprint could hit about 9.3 trillion litres a year, comparable to the basic annual domestic water needs of the 1.3 billion people living in Sub-Saharan Africa. The land footprint of the supporting infrastructure could exceed 14,500 square kilometres, roughly twice the area of metropolitan Jakarta.

The report even prices out a single output: producing one standard-resolution AI image, it estimates, draws about 2.9 watt-hours of electricity, 1.22 grams of CO2-equivalent, 28.6 millilitres of water, and 0.45 square centimetres of land. Multiplied across billions of daily prompts, the abstractions become reservoirs and substations.

"Though often described as weightless and virtual, the reality of AI is profoundly physical," Madani said. "Behind every prompt, image, or video lies a growing infrastructure of energy systems, water withdrawals, land use, mineral extraction, and electronic waste." He called the report "a call to make those hidden environmental costs visible before they become unmanageable."

Will voluntary disclosure work?

The central question hanging over the initiative is whether a non-binding call will change behaviour at companies that have so far disclosed selectively. The track record of voluntary corporate environmental reporting is mixed: it tends to surface flattering metrics and obscure inconvenient ones. The UNU-INWEH report makes precisely this point, warning that "low-carbon" is not automatically "low-water" or "low-land," and that judging AI sustainability through a single metric can hide trade-offs and push environmental burdens onto regions already short of water or arable land.

That is the structural risk. A company can credibly claim a renewable-powered data centre while saying nothing about the millions of litres it evaporates through cooling in a drought-prone county, or the land and mineral supply chains behind its chips. Transparency that is partial can function as cover.

The strain is not hypothetical. Data centres are increasingly sited in water-stressed and grid-constrained regions, where they compete with households and farms for the same aquifers and the same megawatts. As Guterres put it, "AI is also hungry for land, water and power," and the facilities running today's models already consume more electricity than most countries. The 2030 renewable target is ambitious precisely because demand is racing ahead of clean-energy supply; pledging green data centres is easy, but building enough new renewable capacity to cover a doubling of AI energy use in four years is not.

What to watch

The initiative's success will be measured less by speeches than by spreadsheets. Watch whether the largest model developers and cloud providers respond with standardized, third-party-verifiable disclosures, or with selective sustainability reports that emphasize carbon while staying quiet on water and land. Watch, too, whether governments begin converting the UN's voluntary norm into hard regulation, mandating the kind of reporting Guterres has so far only requested.

The deeper test is whether transparency changes anything. Disclosure is a precondition for accountability, not a substitute for it. As Madani framed the stakes, the future of AI should be measured "not only by what machines can do, but by whether humanity can deploy those capabilities within planetary boundaries." The UN has now put a number on the bill. Whether the industry pays it openly is the story to follow.

---

Sources: [EurekAlert! / UN University](https://www.eurekalert.org/news-releases/1133347); [Fortune](https://fortune.com/2026/06/23/un-guterres-ai-climate-impact-disclosure-data-centers/); [Euronews](https://www.euronews.com/2026/06/23/no-more-hidden-costs-un-chief-demands-ai-firms-come-clean-over-environmental-footprint); [Climate Home News](https://www.climatechangenews.com/2026/06/23/un-ai-firms-to-reveal-full-environmental-impacts-data-centres-water-energy/); [UN News](https://news.un.org/en/story/2026/06/1167658); [United Nations University (UNU-INWEH report)](https://unu.edu/inweh/news/environmental-cost-of-AIs-Enrgy-use-carbon-water-and-land-footprints).