At the tenth annual conference of the European Systemic Risk Board (ESRB), ECB President Christine Lagarde identified three critical areas where artificial intelligence poses systemic risks to the financial system. She noted that nearly nine out of ten significant euro area banks currently use generative AI, while seven out of ten EU securities market firms plan to increase their AI investments. Lagarde warned that as AI agents gain more autonomy, they may pursue goals in unintended ways, leading to market misalignment and potential collusion. Additionally, she highlighted that rapid advancements in AI capabilities could shorten the time between a cyber exploit and widespread automated attacks from weeks to hours, challenging traditional defense mechanisms.
The speech emphasized the concentration risk associated with reliance on a few frontier AI models, primarily developed in the United States and China. Lagarde referenced a June incident where a US export-control directive led to an abrupt suspension of access to two advanced models for European users, illustrating how geopolitical tensions can directly impact financial infrastructure. She argued that such dependencies create national security concerns, as loss of access could disrupt trading strategies and vulnerability assessments across the sector. The ESRB Advisory Scientific Committee also cautioned that widespread use of similar models might cause firms to react identically to shocks, reinforcing price movements and amplifying volatility.
Lagarde’s address underscores a fundamental tension in the modernization of financial infrastructure: the efficiency gains from AI adoption are increasingly offset by new vectors of systemic fragility. The shift from algorithmic execution to agentic decision-making introduces opacity that complicates regulatory oversight. When AI systems operate with limited human direction, the potential for undetected misalignment or autonomous collusion creates risks that traditional macroprudential tools may not adequately capture. This evolution demands a reevaluation of how supervisors monitor market behavior, moving beyond transaction-level scrutiny to understanding the underlying logic and incentives of autonomous agents.
Furthermore, the geopolitical dimension of AI dependency presents a strategic vulnerability for Europe’s financial sovereignty. The recent disruption caused by US export controls demonstrates that access to critical defensive and analytical tools is subject to external policy decisions. As financial institutions integrate frontier models into core operations, the distinction between commercial technology supply chains and national security interests blurs. This necessitates urgent investment in domestic AI capabilities and robust international frameworks to prevent fragmentation of the global financial system along geopolitical lines, ensuring that resilience does not become contingent on foreign administrative discretion.
