Marlene Amstad, president of the Swiss Financial Market Supervisory Authority (FINMA), warned that artificial intelligence (AI) is substantially amplifying cyber and national security risks for financial institutions, and urged banks and regulators to adopt new technologies quickly to strengthen their defences. Amstad spoke to Reuters after the first hackathon aimed at developing supervisory tools addressing these issues.
About one hundred policy and technology experts attended the event this week, working on shared tools to help monitor crypto markets and other digital-asset systems. Amstad noted that models designed to spot software vulnerabilities have recently signalled increasingly serious risks, while AI itself raises fresh security and accountability challenges for banks.
"As hackers work faster, banks must adapt and fix vulnerabilities more quickly," Amstad said. She added that, with FINMA's involvement, a dedicated forum has been created within the International Organization of Securities Commissions (IOSCO) to promote AI use among authorities that together cover roughly 95 percent of global financial markets.
Regulators are also exploring how to embed security mechanisms directly into the architectures of digital assets. Experiences with models such as Anthropic's Mythos have already exposed vulnerabilities and operational risks tied to AI. The U.S. government this month ordered Anthropic to suspend exports of its latest models, Mythos and Fable, citing national security concerns.
Meanwhile, Chinese cybersecurity firm 360 Security Technology announced it has developed a domestic alternative to Mythos. Amstad stressed that Switzerland must retain access to the most advanced AI models because the technology can play a key role in hardening systems prior to live deployment.
The core message from FINMA is twofold: regulators and banks must respond to the new threats AI creates, and they should also integrate AI as a defensive tool to keep pace with rapidly evolving attacker techniques.
Why it matters
- The hackathon and the IOSCO forum aim to create shared international supervisory tools to address AI-driven risks.
- Authorities are considering systemic ways to embed security features into digital-asset systems.
- Recent developments — including export limits on Anthropic's models and the emergence of domestic alternatives in China — illustrate how access to AI technology and national security concerns are increasingly intertwined.
Amstad's call to action for the financial sector is clear: shorten vulnerability remediation cycles, deploy advanced supervisory technologies, and strengthen international cooperation so banks can manage the heightened cyber risks amplified by AI.
