Artificial intelligence (AI) companies should disclose how much energy, water and land their systems use, and stop chatbots spreading false information about climate change, under new UN-backed guidance.
The Priorities for Action on AI and Climate Information Integrity was launched on Friday at the COP31 climate summit in Antalya, Turkey.
The plan was drawn up by the Global Initiative for Information Integrity on Climate Change, which is co-chaired by the United Nations, its scientific agency Unesco and Brazil.
“The question isn’t simply whether AI is good or bad for the climate. It’s whether people can rely on what they’re told by AI and about it,” said Charlotte Scaddan, a senior adviser at the UN Department of Global Communications.
The guidance is voluntary, and its impact will depend on whether companies and governments adopt it.
The plan tackles two problems: the reliability of the climate information people get from AI, and the lack of data on AI’s own environmental impact.
The UN’s Independent International Scientific Panel on AI has warned that “producing fluent text is easier than producing factual text”. It has also raised concerns about “sycophancy”, where chatbots reinforce what users already believe.
During floods, storms and other disasters, AI-generated images and videos can spread faster than official warnings, the document says.
At the same time, the public is asked to accept claims about AI’s climate impact without the data needed to judge them. Communities living near AI infrastructure are often “left in the dark” about its environmental effects.
AI developers are urged to measure and publish their energy use and carbon, water and land footprints, using common standards that can be independently checked.
They are also asked to disclose whether their products are sold to the fossil fuel industry for exploration and production, and not to use claims about AI’s climate benefits to offset their reported emissions.
On information, companies should show sources and uncertainty, link to authoritative science and avoid rewarding agreement over accuracy, according to WION. They should test their systems for risks before launch and after major updates, publish how reliable they are in different languages and give independent researchers access to data.
During disasters, they are asked to prioritise official warnings, label AI-generated images and audio, and work with national weather and emergency agencies.
Governments are urged to build information integrity into national AI strategies and climate plans, require AI firms to disclose their environmental impact and keep human oversight where AI is used in early-warning systems. They should also ensure fair payment when original journalism is used to train AI tools.
Researchers are asked to test companies’ claims and build open benchmarks in many languages, while journalists are encouraged to investigate AI’s environmental footprint and verify AI-generated images before publishing them.
A separate charter for the advertising industry, launched the same day, calls for transparency over where ads appear, verification of advertisers on digital platforms and action against greenwashing.
The guidance builds on the Belém Declaration on Information Integrity on Climate Change, launched at COP30 in Brazil last year. It has been signed by 29 countries and the European Union, and remains open for endorsement.
Germany has committed €2m to a Unesco-run fund supporting research, communication strategies and public awareness campaigns.
The initiative will report on progress at the UN’s Global Dialogue on AI Governance and at next year’s COP32 summit. COP31 ends on 20 November.
What it means for Nigeria
More Nigerians are turning to AI chatbots for information, but today’s models work well in only a small fraction of the world’s languages. The call for companies to publish how reliable their systems are across languages, and to work with national weather and emergency agencies during disasters, is directly relevant to a country hit by seasonal flooding, where much public information circulates in the country’s local languages and Pidgin English.
For policymakers, the framework offers a ready checklist as Nigeria shapes its approach to AI governance. For journalists, it is a reminder that verifying AI-generated images and questioning tech companies’ green claims are now part of climate reporting.
Discover more from Pluboard
Subscribe to get the latest posts sent to your email.