Mentions of AI across speeches, minutes and statements grew roughly 11x in five years. Five months into 2026 the count has almost surpassed all mentions of AI in 2025, and the tone is shifting from optimism toward risk and caution.
Click here to see our Central Bank Data
AI was a niche topic in 2020. After ChatGPT's late-2022 launch, mentions tripled in 2023 and kept compounding. The 2026 bar covers only January–May, so the dotted projection shows where a full year lands at the current pace — well past 2025.
Solid = recorded. Dotted = 2026 full-year run-rate (5-month pace ×2.4).
"AI" in the abstract still dominates, but generative AI and large language models are now the second-largest theme — barely a presence before 2023. Governance and policy language remains thin, suggesting institutions are quicker to recognise AI than they are at regulating it.
Mentions by category, by year
Top institutions by mention count
Most mentions frame AI as an opportunity, chiefly productivity and growth. But risk-coded language has climbed steadily: cyber threats, financial-stability concerns, cost implications to natural resources, energy and worries about overvalued AI equities. The gap between the two lines is narrowing.
Mentions coded by stance, by year
% of mentions carrying risk language
Recent excerpts (2025–2026) showing how central banks frame the upside and the downside. Quotes are condensed from the source documents.
Cyber, financial stability, valuation
Productivity, supervision, inclusion
In six years AI went from a curiosity to a standing agenda item and 2026 marks the moment central bankers started weighing the financial-stability risks as much as the productivity gains.