AI does more harm than good to the climate, study shows

A study published in npj Climate Action found that if AI is adopted at the same rate in "dirty" and "green" energy, net annual CO2 emissions increase by 0.47–1.8 gigatons: AI's contribution to increased oil and gas extraction outweighs its potential climate benefits.

Compiled by Sergey KostenchukPublished 2026-08-23Updated 2026-09-18

2026-08-23 Business

💼 AI does more harm than good to the climate, study shows

Researchers at Purdue University assessed the climate impact of AI across the entire energy sector for the first time: in 64 scenarios where AI is adopted at the same rate in "dirty" and "green" energy, net annual CO2 emissions increase by 0.47–1.8 gigatons, or 1–5% of sector emissions.

🌍 AI's impact on the climate was modeled for the first time not through data center energy consumption, but through increased oil and gas extraction. A new category, "enabled emissions" — indirect emissions not covered by ESG frameworks — is emerging for the industry.

👤 The main climate risk of AI is not data center electricity, but accelerated hydrocarbon extraction. This is a reason to check vendors' "green AI" promises against the numbers.

Source 1: https://www.nature.com/articles/s44168-026-00411-0

Source 2: https://www.theguardian.com/technology/2026/aug/11/ai-will-do-more-to-boost-fossil-fuel-production-than-green-energy

Sources