Central banks are turning to artificial intelligence to better understand and handle climate risks that threaten financial stability. These powerful tools help process huge amounts of data and spot patterns that humans might miss. Yet important questions remain about how well the technology works and whether its own environmental costs make sense.
The Growing Need for Better Climate Data
Central banks around the world face a tough challenge. They must assess how climate change could hit banks, insurers, and the wider economy. The biggest problem is data. Many companies do not fully report their emissions, especially scope 3 emissions from supply chains. This leaves huge gaps in what regulators know.
Without good data, it is hard to run proper stress tests or guide monetary policy. Traditional methods take too long and often miss important details hidden in text reports. AI changes this by quickly reading through thousands of documents and pulling out key facts.
This shift could transform how central banks protect the financial system from climate shocks.
Researchers and supervisors have spent years building better ways to measure these risks. The Network for Greening the Financial System brings many central banks together to share approaches. Now AI offers a practical way to turn all that talk into action.
BIS Projects Lead the Way with AI Tools
The Bank for International Settlements has launched two major projects that show what AI can do. Project Gaia uses large language models to search corporate reports and extract climate information. It works with partners including the European Central Bank and Deutsche Bundesbank.
The tool handles different ways companies describe the same things. One firm might call something “net zero commitment” while another uses different words. Gaia makes sense of these differences and creates comparable data. Early tests showed it can pull 20 key indicators from reports of 187 financial institutions across five years.
Project Symbiosis focuses on supply chains, especially in Southeast Asia. It works with the Hong Kong Monetary Authority to help banks understand climate risks from small and medium enterprises. The project uses AI to close information gaps on financed emissions and find real opportunities to cut them.
These BIS efforts matter because they create tools that other central banks can use. They turn messy public reports into structured data that supports better decisions.
AI makes climate risk analysis faster, more transparent, and more consistent across borders.
France Shows How AI Estimates Carbon Emissions
The Banque de France tested AI to fill gaps in corporate emissions data. Their research note from late 2025 found promising results. Machine learning models correctly predicted carbon intensity within a reasonable margin in 69 percent of cases.
The team used a random forest model trained on data from over 7,000 listed firms in 100 countries. It considers factors like sector, company size, investment levels, and whether firms have set carbon reduction targets. The AI approach beat traditional statistical methods, which only hit the target in 39 percent of cases.
Still, the model has limits. It tends to overestimate emissions for cleaner companies and underestimate them for heavy polluters. Experts say human judgment remains important for the most extreme cases.
This work helps the Banque de France in several ways. It supports their efforts to green monetary policy operations and assess risks in the financial system. Better emissions data means better decisions about which bonds to hold and how to manage overall exposure to transition risks.
Other central banks are watching these results closely. The ability to estimate missing data could help smaller regulators who lack resources to chase every company for reports.
Efforts Expand to Supply Chains and Asia
In Vietnam, the State Bank of Vietnam sees AI as a way to improve ESG reporting across the banking sector. Deputy governors have highlighted how the technology can make sustainability data more reliable and easier to produce. This matters in emerging markets where reporting systems are still developing.
AI also shows promise with imagery data. Central banks explore satellite information to track blue carbon stores in coastal ecosystems like mangroves. This gives science-based evidence for physical risks from climate change and nature loss.
Project Symbiosis takes this further by focusing on supply chain transparency. Many banks struggle to know the real environmental impact of loans to smaller companies. AI helps analyze patterns and flag potential risks before they become problems.
These regional efforts matter because climate risk is global. What happens in Asian supply chains affects European banks. Stronger data flows between regions can lead to better global oversight.
Concerns Grow Over AI’s Own Climate Footprint
Not everyone is fully convinced. Training and running large AI models uses significant energy and water. Some researchers worry that the emissions from AI could cancel out the benefits it brings in fighting climate change.
The key is targeted use. Using AI only for specific high-value tasks like emissions estimation or report analysis can deliver net gains. Broad, unfocused applications might create more problems than they solve.
Central banks understand this tension. Many now ask financial institutions to report their own AI-related energy use as part of climate risk assessments. This creates a feedback loop where the technology is used carefully and its impacts are tracked.
The goal is balance. AI should help central banks protect the economy from climate shocks without adding unnecessarily to those same problems.
As these experiments continue, central banks are learning what works best. Success could mean more accurate risk models, better informed policy, and stronger financial stability in a warming world. Failure to manage the technology wisely could create new vulnerabilities.
The coming years will show whether AI becomes a true game changer for climate risk management or just another promising tool with limited real impact. For now, the early results from projects like Gaia and the Banque de France tests offer genuine hope that better data can lead to smarter decisions.
