Central bankers warn agentic AI could roil markets

Central bankers and market supervisors warned that agentic artificial intelligence could amplify market volatility and called for operational guardrails such as circuit breakers or ‘kill switches’ to limit trading during severe model failures.

Bank of England deputy governor Sarah Breeden raised the issue at the European Central Bank’s annual meeting in Sintra, asking whether protections ‘analogous to circuit breakers or kill switches that would limit or stop trading market-wide if faulty AI models cause market meltdown’ are needed.

ECB President Christine Lagarde described AI as a ‘major risk’, noting that the speed and scale of modern models increases risks tied to cybersecurity and data theft and that funding for defensive measures has not been defined.

The head of Britain’s market regulator, Nikhil Rathi, observed that technology now moves in weeks or months and urged new tools and closer collaboration with market participants because the traditional cycle for writing rules is too slow for fast-moving AI developments.

The Bank for International Settlements warned on June 28 that AI ‘exuberance’ could lead to significant financial consequences if a sudden policy shift or market re-pricing produces sharp losses after a period of elevated risk-taking. Tobias Adrian, director of the IMF’s Monetary and Capital Markets Department, warned of a ‘potential maturity mismatch between the duration of the physical assets and the duration of the debt,’ meaning short-term borrowing against long-term AI projects could amplify shocks.

Officials also highlighted the risk that faulty or manipulated models may generate erroneous trading signals or automated activity that cascades through markets. They flagged cyber and data risks as model scale increases and said defensive capabilities and their funding remain unresolved.

Regulators cautioned that strict or slow regulation could encourage firms to relocate to jurisdictions with lighter compliance. They noted that US firms lead in frontier model development and AI investment, while Europe has fewer direct capital channels into AI compared with US equity markets.

Central banks and supervisors proposed a mix of short-term operational measures and longer-term policy work. Ideas raised include market-level circuit breakers or kill switches, supervisory stress testing of AI-driven strategies, enhanced transparency for models that influence markets, and cross-border cooperation on standards and enforcement.

Officials said they will continue to monitor developments in model capabilities, funding patterns and cyber threats while developing targeted interventions to limit market-wide disruption.