As Big Tech faces criticism for the environmental impact of artificial intelligence, companies have said the technology will actually help solve climate change.
But those claims often lack scientific evidence, a new report finds. And when touting the climate benefits of AI, tech companies conflate “traditional AI” with the more environmentally harmful generative AI, a form of “bait-and-switch” that amounts to greenwashing.
The report, commissioned by a group of environmental organizations including Beyond Fossil Fuels, Friends of the Earth, and Stand.earth, analyzed 154 statements from tech companies, including those from Google and Microsoft, which purported that AI will have a “net climate benefit.”
Most of those comments relate to “traditional” AI, the analysis found, which has a smaller environmental footprint than the generative AI tools that are spurring a boom in data centers.
Tech companies, however, tend to lump these technologies together, the report says, blurring the differences and presenting the climate benefits and environmental harms as a “package deal.”
But whether those climate benefits are real is also unclear. Only 26% of those statements cite published academic papers, the research found, and 36% don’t cite any evidence at all.
Of the remaining statements, 29% cited corporate publications—the majority of which did not include peer-reviewed or published academic work—and 8% cited media, NGOs, or unpublished academic papers.
Questionable AI evidence
The rapid expansion of AI has come under fire for its potential environmental harms. Reports on generative AI’s climate impact vary, but the tech has been linked to intense energy and water use.
Tech companies have justified their AI expansion by pointing to AI’s climate benefits. One of the most widespread claims is that AI could help mitigate 5% to 10% of global greenhouse gas emissions by 2030.
Google has repeated that statistic, including in its 2024 environmental report. That figure, however, comes from a 2021 blog post by consulting firm BCG which attributes it to the firm’s own “experience with clients.”
“This questionable extrapolation of massive global climate benefits justified on seemingly anecdotal evidence was the first clear instance of what has become a longer-term trend of overstating the climate benefits of AI,” the report reads.
In reality, the International Energy Agency (IEA) projects total data center consumption, driven by AI, will double by 2030—and Bloomberg New Energy Finance estimates that this increase will grow total global power sector emissions by 10% over the coming decade.
In another example, Google’s 2025 environmental report said that rooftop solar power installations assisted by an AI mapping tool would “help enable partners to reduce around 6 million metric tons of lifetime GHG emissions.”
Google also said that figure would be “around 6,000 times greater” than the service’s operation in 2024.
But the Beyond Fossil Fuels report says that Google’s footnotes reveal that the 6 million figure is an estimate of the total emissions avoided by rooftop solar because they produce low-emissions energy, not the additional reductions from the AI mapping tool.
This detail could “create the impression,” the report notes, that the climate benefits are attributed to the AI tool.
In response to a request for comment, a Google spokesperson told Fast Company that it stood by its methodology, “which is grounded in the best available science. And we are transparent in sharing the principles and methodology that guide it.” (That methodology does not mention AI.)
Microsoft, also cited in the report, declined to comment.
What even counts as ‘AI’?
To many, any mention of “AI” has become synonymous with generative AI—examples of which include large language models like Claude, ChatGPT, and Copilot, and image or video generating services like Midjourney and Sora.
But not all AI is “generative.” Traditional AI, an umbrella term that covers subsets like machine learning, has been powering all sorts of technology for years, from search engines and recommendation algorithms to medical imaging.
Generative AI consumes more energy and is associated with more emissions than traditional AI. When tech companies talk about AI’s climate benefits, though, they can conflate the two terms, or position them like a “package deal.”
Most AI climate benefits will come from traditional AI, the report found. In its analysis, the researchers said that “at no point” did they “uncover examples in which consumer generative systems were leading to a material, verifiable, and substantial level of emissions reductions.”
So climate benefit claims are attributed to traditional AI, but the majority of energy consumption comes from generative AI.
The surge in data center demand is largely driven by the exploding demand for generative AI. Those data centers are also directly spurring more natural gas in the U.S.
The confusion between these terms matters, the report says, because it amounts to a “bait-and-switch” type of greenwashing: Tech companies are justifying their data center expansion by touting AI’s climate benefits, though most of those data centers “will not be processing climate-beneficial computation on their servers.
“Big Tech’s AI hype is distracting users from the rapid and dangerous expansion of giant, energy and water-intensive data centers, while the tech industry’s huge energy demands are throwing the fossil fuel industry a lifeline,” Jill McArdle, international corporate campaigner from Beyond Fossil Fuels, said in a statement.
“There is simply no evidence that AI will help the climate more than it will harm it,” she added. “We cannot bet the climate on these baseless claims.”