Swiss regulator urges banks to adopt AI tools against cyber threats

Marlene Amstad, president of FINMA, said banks and supervisors need their own AI tools as cyberattacks and market threats accelerate, or risk falling behind attackers who can probe weaknesses faster than human teams can respond. She said Switzerland must retain access to the most advanced AI models, arguing that AI will be essential to hardening systems before they are put into use. An initial hackathon aimed at helping supervisors build practical oversight tools produced systems that could monitor crypto markets and spot vulnerabilities more quickly. AI can help banks automate work, but it also gives hackers faster ways to launch phishing campaigns, generate malicious code and exploit system gaps before defenders react.

Finance runs on speed, data and trust, making it unusually exposed. AI compresses market timeframes so price discovery can happen in milliseconds and credit decisions in seconds, shrinking the window for regulators to catch errors before they spread. FINMA is already developing AI-powered supervisory tools based on supervised machine learning models and network analysis, with near real-time crypto monitoring singled out as a key use case.

The International Organization of Securities Commissions conducted its first SupTech survey in early 2025 across all IOSCO regions, with 49 authorities taking part. Supervisory technology is no longer experimental and is being folded into core supervisory work. IOSCO has more than 200 members from more than 130 jurisdictions, and its members regulate 99% of the world’s financial markets, giving the shift broad reach beyond Switzerland.

In May 2025, Amstad told the IOSCO Presidents Committee that supervisors should benefit from advances in technology, use new supervisory instruments and strengthen international cooperation. In October 2025, IOSCO launched the SupTech Forum, and Amstad chairs it.