AI-simulated fed meeting shows political pressure divides board
A new study has revealed how political pressure can sway U.S. Federal Reserve decision-making, even in an AI-simulated environment.
Researchers at George Washington University recreated a Federal Open Market Committee (FOMC) meeting using AI agents trained on real policymakers’ histories, speeches, and policy stances. The simulation processed live economic data and news to reach decisions, News.Az reports, citing Reuters.
The findings showed that when political pressure was introduced, board members—represented by AI—became polarized, with dissent more frequent. The paper, by academics Sophia Kazinnik and Tara Sinclair, concluded that the Fed is “only partially insulated from politics,” and that outside scrutiny can influence internal deliberations.
The experiment replicated the July 2025 FOMC meeting. While central banks are unlikely to let AI set policy, many are already using the technology to enhance analysis and operations. The Federal Reserve has experimented with generative AI to study meeting minutes, the European Central Bank has deployed machine learning to forecast euro-area inflation, and the Bank of Japan is using AI to track shifts in price drivers. Australia’s central bank is also testing AI tools to improve research efficiency.
The Bank for International Settlements has highlighted AI’s growing strategic role in central banking but stressed the importance of strong governance and reliable data as adoption expands.
Similar news
