Federal Reserve Board’s Cook says the FOMC is not using AI to set policy and sets out a framework for responsible adoption

In a speech, Federal Reserve Board Governor Lisa D. Cook set out her view of artificial intelligence as an emerging general-purpose technology with the potential to affect both maximum employment and price stability, while describing how the Federal Reserve is studying and selectively using AI internally. She drew a clear line between operational use and monetary policymaking, stating that the Federal Open Market Committee is not using AI in developing or setting policy, but that AI tools may support staff work such as writing, coding, and research. Cook outlined four principles for responsible AI adoption: strong governance and risk management with “humans in the loop,” safeguards for privacy, cybersecurity, and confidential information, investment in education and training, and a culture of empowerment and experimentation in controlled environments with the ability to stop projects that do not meet standards. She highlighted a range of Federal Reserve research efforts using large language models and machine learning, including work on interpreting FOMC minutes, measuring private-sector AI research activity via earnings calls, simulating the Survey of Professional Forecasters, forecasting financial crises from unstructured text, and producing real-time layoff indicators from filings, alongside findings that point to limitations for real-time use such as look-ahead bias and challenges with vintage data. On the broader economy, she argued that adoption frictions include workforce training needs, the time required for organizational learning and diffusion, and a rational incentive to be selective while the technology changes rapidly; for monetary policy, she emphasized uncertainty around net effects, with AI potentially raising productivity and lowering inflationary pressures over time but also possibly boosting prices in the interim if adoption drives a surge in aggregate investment.