Climate Law: New AI Tool Produces Reports on Fossil Fuel Company Activities 

March 25, 2025 -- A new open access AI tool created by researchers at the University of Oxford’s Sustainable Law Programme produces reports on the historical activities of fossil fuel companies in response to user prompts.

The Climate Accountability Research Assistant, or CLARA, is a Large Language Model (LLM) agent that draws on over 14,000 pages of historical fossil fuel industry documents to produce concise reports with citations in a matter of minutes. It can answer questions such as “When did fossil fuel companies become aware they were harming the environment?” or “How have fossil fuel companies contributed to climate denial?”

“Louis Brandeis once said the light of day is the best disinfectant. CLARA will help researchers, journalists and lawyers alike shine light on the past actions and statements of fossil fuel companies with speed and accuracy,” says Dr Benjamin Franta, Associate Professor and Senior Research Fellow in Climate Litigation at the Oxford Sustainable Law Programme and the founding head of the Climate Litigation Lab.

“To our knowledge, it is the only tool of its kind, and we believe it is a major advance in the use of AI for corporate accountability, legal action, journalism, and research in this space,” continues Dr Franta.

“Efforts to hold corporations accountable for their actions have often been held back by the sheer amount of information that must be sifted through in the legal research and discovery phases. Our hope is that CLARA can make this process more efficient,” says Data Scientist Meghana Patakota, who developed the tool alongside Research Engineer Jake Rutherford.

Ensuring the reliability of CLARA's output was crucial, emphasizes Rutherford: "We needed CLARA's responses to be accurate, correct, and verifiable, making it suitable for rigorous academic research and robust legal case-building."

Hallucinations are an inherent limitation of large language models, and so Patakota and Rutherford adopted a methodology explicitly designed to minimize these errors. CLARA has been programmed to:

• Uncover relevant information and effectively reason about entities such as individuals, companies, events, and their interconnections
• Analyse entire documents when necessary, so as to substantiate its findings with complete contextual understanding
• Iteratively refine its research strategy in response to emerging insights during its use.

Together, these attributes enhance the effectiveness and trustworthiness of the underlying LLM. “It’s like the difference between telling somebody the information and asking them to recall it from memory, or asking them to look at the sources and repeat the information back,” explains Rutherford.

 “CLARA’s ability to produce information that's verifiable and cited is a gamechanger for AI research tools,” says Dr Franta. “Our hope is that it becomes an essential research partner for uncovering the fossil fuel industry's secrets and that it can empower current and future climate cases. I also hope to see this approach being applied to other areas in need of corporate accountability, including to address plastics and other industrial pollutants.”

CLARA is available for open access use here: https://clara-research.com.

About the Smith School of Enterprise and the Environment

The Smith School of Enterprise and the Environment at the University of Oxford equips enterprise to achieve net zero emissions and the sustainable development goals, through world-leading research, teaching and partnerships.  Visit https://www.smithschool.ox.ac.uk.

Source: Smith School of Enterprise and the Environment