The International Monetary Fund is a multilateral financial institution established in 1944, in the aftermath of the Great Depression, by 44 founding member countries; it now has 191 member countries. It fosters international financial stability through policy advice and surveillance of economic and financial developments, lending to member countries, and capacity development (technical assistance and training) (Source: imf.org). Its AI work began with labour-market and capital-market analysis in 2024, and in 2026 it began framing frontier-AI capability as a systemic-financial-risk question.
Labour-market and preparedness analysis
In a January 14, 2024 blog accompanying the staff discussion note Gen-AI: Artificial Intelligence and the Future of Work, Managing Director Kristalina Georgieva reported the staff finding that "almost 40 percent of global employment is exposed to AI." Exposure was estimated at about 60% of jobs in advanced economies, of which roughly half may benefit from AI integration while for the other half AI may execute key tasks and lower labour demand, and at 40% in emerging markets and 26% in low-income countries. Georgieva wrote that "in most scenarios, AI will likely worsen overall inequality" and called for social safety nets and retraining. The same work introduced an IMF AI Preparedness Index measuring digital infrastructure, human-capital and labour-market policies, innovation and economic integration, and regulation and ethics across 125 countries, on which Singapore, the United States and Denmark scored highest (Source: imf.org). See AI Labor Disruption.
AI in capital markets (October 2024)
Chapter 3 of the October 2024 Global Financial Stability Report, "Advances in Artificial Intelligence: Implications for Capital Market Activities," drew on analytical work and outreach to market participants and regulators. It found that AI adoption in capital markets was likely to increase significantly and that AI might reduce some financial-stability risks through better risk management, deeper liquidity and improved market monitoring. It identified new risks: increased market speed and volatility under stress, greater opacity and monitoring challenges for non-bank financial institutions, operational risk from reliance on a few key third-party AI-service providers, and increased cyber and market-manipulation risks (Source: imf.org).
AI policy activity
A May 7, 2026 IMF financial-stability blog, "Financial Stability Risks Mount as Artificial Intelligence Fuels Cyberattacks," by Tobias Adrian, Tamas Gaidosch and Rangachary Ravikumar, named specific frontier-AI capability — the Mythos preview and GPT-5.5-Cyber — as systemic-financial-risk vectors. It is the first IMF designation of a specific frontier model in those terms and the anchor source for AI Macro-Prudential Policy. Until May 2026 the IMF treated AI as an operational-risk category, that is, something a bank uses internally; the May 7 blog instead treated AI capability itself as a systemic-risk category. In the blog's own words, Anthropic's controlled release of Claude Mythos Preview "underscored how quickly risks are increasing" and "foreshadows how fast‑moving, AI‑driven cyber risks could destabilize the financial system if not managed carefully"; it described OpenAI's restricted cyber version of GPT‑5.5 as an approach that "assumes vulnerabilities and attacks will grow, and emphasizes equipping defenders." It called cyber stress testing, scenario analysis and board-level oversight of cyber risk "indispensable components of financial stability frameworks," and warned that emerging and developing economies "may be disproportionately exposed" (Source: imf.org). ECB President Christine Lagarde said on May 8, 2026 that the ECB was studying defences against Mythos-powered attacks (Source: reuters.com); see AI Macro-Prudential Policy for the wider propagation. The blog likened the shift to the way collateralized debt obligations moved from an obscure financial instrument to a stress-test category after the 2008 financial crisis.
The IMF blog followed earlier framing from G7 central banks. Per a New York Times report of April 22, the Governor of the Bank of England warned publicly that Anthropic may have "crack[ed] the whole cyber-risk world open," roughly 2 weeks before the IMF designation. The same New York Times report stated that the European Central Bank had begun "quietly questioning banks about their defenses" in the same window.
Whether the macro-prudential AI framing becomes operative depends on G7 central banks — the Federal Reserve (Fed), ECB, Bank of England (BoE), Bank of Japan (BoJ), and Bank of Canada (BoC) — citing the IMF designation in their own stress-test scoping or supervisory guidance. As of the May 2026 record, zero of the 5 had publicly done so. The Bank of England's stress-test scoping work is described as the most likely first mover, given its Governor's public framing ahead of the IMF designation.
AI investment and leverage
IMF officials have also addressed the financial risks of the AI investment boom. At the European Central Bank's symposium in Sintra on July 1, 2026, the IMF's Tobias Adrian called leverage "on both sides" (hyperscaler debt issuance and investor leverage) worrisome for financial stability (Source: livemint.com). An essay in The Atlantic cites Gita Gopinath, identified there as of the IMF, estimating that an AI crash could wipe out roughly $35 trillion in global wealth (Even Silicon Valley Says That AI Is a Bubble — Lila Shroff (The Atlantic, March 2026)). See AI Bubble Debate.
Background note to EU finance ministers (September 2026)
In a background note prepared for the informal meeting of EU finance ministers in Dublin on September 18 and 19, 2026, the IMF estimated that AI could raise European productivity by about 1% over five years while risking wider inequality, strained power networks and deeper dependence on foreign technology. The paper put the share of workers in advanced European economies holding occupations highly exposed to AI at around 60%, and European data centres' consumption at roughly 3% of the continent's electricity, naming Frankfurt, London, Amsterdam, Paris and Dublin among the most exposed clusters. Its recommendations were to complete the EU single market, invest in cross-border grid infrastructure and deepen energy-market integration. It warned that US and Chinese dominance of model development risks creating another strategic dependency for Europe (Source: reuters.com).
Relationships
- anchors: AI Macro-Prudential Policy.
- related: Claude Mythos Preview, AI and Cybersecurity, AI Bubble Debate, AI Labor Disruption, Even Silicon Valley Says That AI Is a Bubble — Lila Shroff (The Atlantic, March 2026).
- related: Bank of England (BoE), European Central Bank (ECB), Federal Reserve, Financial Stability Board (FSB).
Provenance note: The May 7, 2026 IMF blog is queued for foundational ingest (
INGEST-imf-ai-cyberattacks-financial-stability-blog-2026.md); citations to it should upgrade to itssources/page once ingested. Further ingest candidates: IMF Global Financial Stability Reports and Article IV consultations citing AI.