An investigative report by Reuters showed that 15 international environmental organizations rely on artificial intelligence to monitor wildlife and protect forests, while simultaneously demanding stricter controls on energy and water consumption in massive data centers ahead of the COP31 climate conference.
Reuters investigation into 15 environmental organizations using artificial intelligence
Elephant rescue experiment in Kenya and analysis of 200,000 aerial images
Jane Goodall Institute and the digitization of historical chimpanzee records
Wildfire monitoring and reliance on lightweight local computing
The equation of data center energy and water consumption ahead of COP31
Frequently Asked Questions
Reuters investigation into 15 environmental organizations using artificial intelligence
All 15 environmental organizations and conservation institutes interviewed by Reuters as part of an investigative report prepared by journalist Valerie Volcovici and published on November 7, 2026 at 06:07 UTC (10:07 Gulf Standard Time) confirmed that they use artificial intelligence technologies in their fieldwork, while simultaneously expressing clear concern about the environmental costs of the data centers powering many AI applications.
This investigation provides documented, new practical examples of artificial intelligence applications beyond the scope of general chatbots, though it does not prove that all environmental organizations worldwide are adopting the same approach, given that the interviewed group does not represent a comprehensive statistical survey of every global environmental organization.
Elephant rescue experiment in Kenya and analysis of 200,000 aerial images
Save the Elephants in Kenya uses AI-enhanced image analysis technologies to conduct large-scale wildlife surveys. Researchers reported that they were able to review and examine approximately 200,000 aerial images within just a few hours, compared to the previous manual screening process which took 9 people 9 continuous months.
The organization explained that automated analysis revealed animals that were not detected during manual counting, including significant proportions of elephants, giraffes, zebras, and oryxes. It is worth noting that these figures represent the results of cases reported by the organization itself, rather than an independent standard for error rates that can be generalized to all species and terrains; environmental photography can be affected by shadows, obstacles, land topography, and varying animal sizes, making the most useful metric verified detection performance on entirely brand-new images with independent human auditing.
Jane Goodall Institute and the digitization of historical chimpanzee records
In a related context, the Jane Goodall Institute is applying artificial intelligence technologies to digitize and analyze handwritten notes on chimpanzee behavior compiled over many years, including materials documented in local languages. According to the revised plan, the process previously expected to take between 13 and 15 years may be shortened to just about two years.
This figure represents a future time estimate under the new plan rather than a completed task that took 15 years and was actually compressed into two verified years on the ground; handwriting recognition and automated transcription require careful human review, especially when dealing with specialized scientific notes or local language variations that algorithms may struggle to interpret accurately.
Wildfire monitoring and reliance on lightweight local computing
Other uses reported in the investigation include identifying potential wildfire threats by analyzing satellite imagery, examining marine fishing activities, and conducting certain fish-related video analysis locally on edge devices rather than relying on computational inference via distant data centers.
Prominent organizations such as Conservation International, the World Resources Institute, The Nature Conservancy, and Greenpeace reviewed various practices and concerns; some groups avoid energy-intensive applications, reuse pre-stored model outputs, or choose smaller computing workflows when appropriate.
The equation of data center energy and water consumption ahead of COP31
This trend reflects a genuine environmental tension; data center demand for electricity, water, and equipment could surge with the expansion of artificial intelligence, noting that the International Energy Agency’s projections for future consumption remain forward-looking forecasts rather than measured results for 2030. The actual environmental footprint of any technical workflow depends on model size, hardware used, electricity source, geographic location, request frequency, and whether computing occurs via a remote server or local device. Therefore, simplistic assertions that all artificial intelligence either harms or benefits the climate ignore these core distinctions.
This report arrives ahead of the COP31 climate conference to demonstrate that environmental organizations can simultaneously employ artificial intelligence to serve their missions while objecting to the scale or design of infrastructure expansion, whereas upcoming required evidence includes conducting independent impact assessments, publicly documenting conservation results, and accounting for consumed resources in a comparable manner.
Frequently Asked Questions
Question: How did artificial intelligence help Save the Elephants in Kenya?
Answer: It enabled researchers to examine about 200,000 aerial wildlife images in hours instead of manual labor that took 9 people 9 months, and helped detect elephants, giraffes, zebras, and oryxes.
Question: What is the Jane Goodall Institute’s AI-backed project?
Answer: The institute is working to digitize and analyze archived handwritten notes on chimpanzees documented in local languages, with estimates of reducing expected completion time from 13 to 15 years to about two years.
Question: How do environmental organizations reduce the carbon footprint of their AI use?
Answer: By avoiding energy-intensive models, reusing pre-stored outputs, and running video analytics locally on field devices instead of sending them to distant data centers.
Question: What determines the actual environmental impact of any artificial intelligence system?
Answer: The impact depends on model size, hardware type, electricity generation source, geographic location, frequency of repeated requests, and whether processing happens in the cloud or on local devices.