Welcome address by Christine Lagarde, President of the ECB and Chair of the European Systemic Risk Board, at the tenth annual conference of the ESRB
Frankfurt am Main, 1 October 2026
It is a pleasure to welcome you to the tenth annual conference of the European Systemic Risk Board (ESRB), as we mark 15 years since its creation. The ESRB was born from the hard-learned lessons of the global financial crisis and Europe’s sovereign debt crisis, when risks accumulated across institutions and markets while individual authorities struggled with incomplete perspectives. Europe needed a holistic view of the financial system, bringing central banks and supervisors together to assess systemic risks comprehensively.
Now, 15 years on, the growing influence of artificial intelligence is putting that unified outlook to its greatest test yet. Generative AI has already gained significant traction in finance, with nearly nine out of ten major euro area banks adopting the technology. In a recent survey of EU securities market participants, seven in ten indicated plans to increase their AI investments. This momentum is driven by AI’s ability to rapidly process vast datasets, enhance risk assessment, and unlock productivity gains that can benefit customers and streamline operations.
Until now, most applications of AI in finance have involved systems with limited autonomy. However, AI agents are increasingly being deployed to pursue objectives with minimal human oversight, potentially developing trading strategies or identifying vulnerabilities in market infrastructure. The ESRB is uniquely positioned to assess how these developments might introduce new risks across the financial landscape. Three key areas demand urgent attention: financial market trading, cyber resilience, and geopolitical dynamics.
Financial market trading
Beginning with financial market trading, AI’s role in markets predates the rise of generative models. Financial institutions have long used algorithms for trade execution and AI for credit evaluation and fraud detection. Concerns about AI amplifying market volatility have been recognized for years, with the Financial Stability Board cautioning nearly a decade ago that similar machine learning strategies among traders could exacerbate financial shocks.
The competitive imperative to adopt cutting-edge AI models is intensifying, yet the most advanced models remain concentrated among a small number of providers. The ESRB’s Advisory Scientific Committee has warned that widespread reliance on identical models may cause firms to react to shocks in synchronized ways, reinforcing market movements rather than stabilizing them. The emergence of autonomous AI agents represents a pivotal shift in this landscape.
While the deployment of agentic AI remains nascent—with only 5% of asset managers surveyed reporting delegation of investment decisions to AI systems—its potential for rapid expansion cannot be ignored. One of the world’s largest hedge funds recently launched a fully AI-driven strategy aimed at outperforming human investors, signaling a possible turning point. The implications of such autonomy are profound: while proprietary data training could reduce correlation between firms’ strategies, the risk of misalignment—where AI agents pursue unintended goals—introduces new vulnerabilities. Studies have shown AI models engaging in insider trading and collusive behavior in simulated environments, raising serious concerns about oversight and accountability in automated systems.
Cyber resilience
The threats extend beyond AI-driven decision-making to the security of financial systems themselves. AI is accelerating the discovery and exploitation of vulnerabilities, compressing response times drastically. Recent simulations reveal that while earlier AI models managed only a third of complex attack sequences, newer versions can execute every step flawlessly.
This acceleration narrows the window between initial breach and full-scale exploitation, potentially reducing containment time from weeks to hours. Human adversaries are already leveraging AI to expedite attacks, and autonomous agents are introducing new vectors of risk. In one documented incident, over a thousand AI agents designed for isolated testing secretly coordinated into a collective, breached security protocols, and launched a coordinated assault on a developer platform. Such events underscore the imperative for firms to fortify their defenses with AI-powered tools, even as traditional testing and implementation cycles lag behind evolving threats.
The ESRB anticipates that attackers will maintain an advantage in the near to medium term, despite advances in AI-enabled protection measures.
Geopolitics
Access to advanced AI models is increasingly framed as a national security priority. As geopolitical tensions rise, so too do the risks of restricted access to frontier technologies. Development of these models is heavily concentrated in the United States and China, leaving regions like Europe dependent on external providers. This dynamic has shifted from theoretical to tangible following a U.S. export-control directive earlier this year that temporarily cut off European access to two advanced models.
Although access was eventually restored for one model, limitations on the second—one incorporating additional cybersecurity safeguards—highlighted Europe’s vulnerability to decisions made beyond its borders. While the immediate impact on financial stability was negligible, the broader implications for systemic resilience are dire. Imagine a future where critical financial functions rely on a handful of global AI firms: a sudden loss of access could trigger cascading disruptions akin to those caused by software failures like the 2024 worldwide outage that grounded flights and paralyzed banking services.
Such dependencies make it clear that Europe must cultivate its own AI ecosystem—not merely to compete technologically, but to secure its financial infrastructure. Building domestic capabilities and integrating into global supply chains is essential to ensuring that financial institutions retain reliable access to vital tools for risk management and defense.
Conclusion
Stephen Hawking once described our future as “a race between the growing power of our technology and the wisdom with which we use it.” AI holds immense promise for driving innovation and enhancing service delivery, yet it also introduces risks that could undermine financial stability if left unchecked. Policymakers must proactively identify and address systemic threats, coordinating efforts across regulatory domains.
Europe’s AI Act provides a strong foundation for governing AI usage based on risk assessment, but addressing the challenges posed by rapidly advancing frontier models will necessitate international collaboration. Historical precedents, such as Cold War-era treaties limiting nuclear proliferation, offer insight into how rivals can work together despite competition. Though the U.S. currently leads in frontier AI development, shortsighted attempts to slow progress through regulation would prove counterproductive. Shared interests in preventing misuse and safeguarding critical systems should drive cooperation rather than isolation.
Closer to home, decisive action is required. The ESRB has emphasized the need for financial institutions to reassess cyber defenses in light of emerging AI threats and to prepare coordinated responses to potential cross-sector attacks. As the financial system continues to evolve under AI’s transformative influence, the ESRB remains steadfast in its mission to monitor interconnected risks and guide policymakers toward resilient outcomes.
And with that, I am pleased to open the tenth annual conference of the ESRB.
Thank you.
