THE LEVERS
What can be done
AI is being built into the search for, drilling of, and production of more oil and gas, and the emissions it enables are not yet captured by reporting standards, ratings methods, or governance. Stopping this practice starts with making those emissions counted, disclosed, and regulated.
01
Lawmakers and standard setters
Decide what gets counted.
Accounting standards
The GHG Protocol is the accounting standard that reporting frameworks globally draw on. Its proposed, optional Scope 3 category 16 for facilitated emissions could make the emissions enabled by technology companies' AI tools a defined, disclosed category, if it includes a clear calculation method.
Reporting frameworks
SBTi, CDP, and the EU’s CSRD do not currently count the emissions that technology companies’ products enable, even when it is the product’s indented purpose; each could ask companies to account for these emissions, with this change reaching many companies at once.
EU AI Act
The EU AI Act names environmental protection as one of its purposes, but its list of high-risk uses in Annex III has no environmental category and does not name fossil fuel operations. Adding them, through a Commission amendment within an existing area or a change to the Act itself, would bring AI systems that expand fossil fuel production under its obligations.
US legislation
The Artificial Intelligence Environmental Impacts Act of 2026, introduced in the Senate in June 2026, focuses almost exclusively on datacenters, and is a natural vehicle for adding disclosure of AI used to expand fossil fuel production, whose added emissions peer-reviewed research puts at 3.3 to 13.3 times the IEA's estimate of current datacenter emissions.
02
Investors and ratings providers
Complete the scores and move capital.
Shareholder resolutions
Shareholders can file resolutions asking companies to report on the emissions its products enable; even when the vote is advisory, it puts the question on the record in front of every large shareholder.
ESG ratings methodologies
None of five major ESG ratings providers (MSCI, S&P Global Sustainable1, Morningstar Sustainalytics, LSEG, or ISS ESG) currently defines or scores the emissions a technology company’s products enable with fossil fuel customers; adding this category would let that enablement show up in how companies are rated and screened.
Institutional investor frameworks
IIGCC and PRI guidance on AI focuses on datacenter energy and water, and extending it to enabled emissions would give asset owners a ready set of questions for the companies they hold.
03
Companies, employees, and customers
Change decisions where they are made.
Internal advocacy
Employees can raise the issue directly with company leadership and sustainability teams, where decisions about which customers and uses to pursue are made.
Public commitments
Companies can make specific, verifiable commitments not to develop custom AI tools for fossil fuel expansion.
Procurement and contract terms
Any organization that buys AI or cloud services, whether a city government or a small business, can make a vendor’s work for fossil fuel producers part of how it evaluates suppliers, the way they screen for issues like forced labor.
04
Courts and communities
Put existing law to work.
Misleading climate claims
A company’s statements about its own climate role, such as Microsoft’s claim to "ensure the technology we build benefits... the planet itself" or how “At Microsoft, we’re committed to helping build the societal conditions to support a net zero economy,” or Microsoft’s chief sustainability officer saying “The systems we build to support the future must also support the long-term health of the planet and the communities we serve,” can be tested against what its business does. A Paris court found TotalEnergies’ claim to be a “major player in the energy transition” likely to mislead consumers in 2025, and the same test applies to a technology company that makes that claim while selling AI to expand fossil fuel production.
Misleading omissions
Marketing that promotes a company’s sustainability projects can be tested for what it leaves out. UK advertising regulators banned Shell ads in 2023 for giving the impression that much of its business was low-carbon, and the same test applies to technology companies that promote their AI for sustainability while leaving out their fossil fuel portfolio. For example, Microsoft’s 2025 Impact Summary says AI “is being used to reduce water loss in cities, strengthen global early warning systems, optimize energy grids and permitting, and streamline corporate disclosures,” and omits any mention of how its AI is being used to expand every stage of fossil fuel development, at a scale that dwarfs its sustainability applications.
Human rights frameworks
Communities harmed by fossil fuel expansion can document the role AI played, which grounds their claims in existing human rights frameworks rather than in new law.
05
Civil society
Build the evidence and the pressure.
Coalition building
A broad coalition of climate, digital rights, and consumer groups can carry the issue into campaigns that already have audiences.
Media and investigations
Journalists and watchdog groups can investigate what technology companies are building for fossil fuel producers and report on it alongside AI’s datacenter energy use, giving the public, regulators, and investors the full picture to act on.
Peer-reviewed research
Independent, peer-reviewed research makes enabled emissions a quantifiable, citable category.
Join the work
If you are a lawmaker, investor, lawyer, journalist, employee, or member of the public who wants to get involved, we would like to hear from you.