Lagarde Warns AI Is a Financial‑Stability Risk, Urges Focus on Trading, Cyber and Europe’s Provider Dependence

Christine Lagarde warned Thursday that artificial intelligence is becoming a financial-stability issue, not just a technology story, and said regulators should concentrate on three risks in particular: trading, cyber resilience and Europe’s dependence on a small number of frontier AI providers.

“I see three areas that call for particular attention,” Lagarde said in Frankfurt, where the European Central Bank published her speech, “Where AI risks meet,” on Oct. 1.

The remarks came in Lagarde’s role as chair of the European Systemic Risk Board, the European Union’s macroprudential watchdog for system-wide financial risks, and as the welcome address at the ESRB’s 10th annual conference. They were a warning about oversight priorities, not an ECB monetary-policy announcement and not a signal on interest rates.

Lagarde said the spread of frontier, generative and agentic AI is testing the ESRB’s mandate because the technology can create risks that cut across firms and markets. In trading, she said, the growing use of similar models across financial companies could make behavior more correlated and amplify market swings. She also pointed to research on algorithmic collusion and on large language models showing strategic deception.

Her second concern was cyber resilience. Lagarde referred to the ESRB’s July 7 warning that frontier AI models could strain the financial system’s defenses against cyberattacks. She said ESRB analysis shows frontier models are improving rapidly at multi-step attack simulations, raising concerns that the same systems being adopted across finance could also strengthen attackers’ capabilities.

The third risk was geopolitical: Europe’s access to the most advanced AI models. Lagarde said the region is vulnerable if critical tools are controlled by a handful of providers and access can be restricted by foreign-policy decisions outside Europe.

The warning comes as AI adoption in finance is already widespread. “Nearly nine out of ten significant euro area banks use” generative AI, Lagarde said, citing ECB supervisory material. She also cited a European Securities and Markets Authority survey finding that seven out of 10 respondents expected to increase AI investment between 2025 and 2027.

At the same time, Lagarde suggested the most autonomous uses of AI in investing are still limited. A 2026 survey she cited found only about 5% of asset managers currently give AI autonomous or semi-autonomous authority over investment recommendations or trades. That suggests the technology is already embedded in the sector even before the most aggressive forms of delegation become common.

On geopolitical concentration, Lagarde pointed to a June 2026 U.S. export-control directive that led one provider to suspend access to two advanced AI models, abruptly cutting off European users. She used the episode as an example of how quickly access to critical models could be disrupted, with potential consequences for European financial firms that rely on them.

Lagarde said Europe already has a rulebook in place through the AI Act, the bloc’s risk-based law for AI systems, and DORA, the framework for digital operational resilience in financial services. But she argued that those tools will not be enough on their own. Frontier models, she said, will also require global cooperation, while Europe should build up its own capabilities to reduce strategic dependence.

The speech builds on the ESRB’s earlier cyber warning and broadens it into a wider financial-stability case for paying closer attention to AI. Lagarde did not announce new regulation, new ESRB measures or any change in ECB policy. Instead, she set out a message that is likely to shape how European regulators think about AI from here: as a potential source of systemic risk, not simply a productivity tool.