FINMA Publishes Guidelines on the Use of Artificial Intelligence

The Swiss Financial Market Supervisory Authority FINMA publishes guidelines on the use of artificial intelligence in the financial sector. With this, Switzerland sets clear regulatory frameworks for AI deployment in supervised institutions.
What the FINMA Guidelines Regulate
The FINMA guidelines address key governance questions around AI use:
Accountability: Supervised institutions must define clear responsibilities for AI systems. Delegating to AI does not exempt from accountability.
Transparency: AI-supported decisions must be traceable. Black-box models without explainability must be critically scrutinized.
Risk management: AI systems must be integrated into existing risk management frameworks. Model risk management becomes the standard.
Data protection: The use of customer data for AI training is subject to strict requirements – especially in the context of the Swiss Data Protection Act (nDSG).
Impact on the Financial Sector
For banks, insurers, and other supervised institutions, the guidelines mean concrete action requirements:
- Documentation of all deployed AI systems
- Regular review and validation of AI models
- Training of employees in the use of AI
- Clear escalation processes for AI errors
Transferability to Other Industries
Even though the FINMA guidelines primarily address the financial sector, they set a standard that is relevant across industries. Companies that implement responsible AI governance today will be better prepared for upcoming regulations tomorrow. Why many companies fail to see results from their AI investments despite governance failures is explored in our post on AI ROI.
How Business IT Partners Helps
We support companies in the regulation-compliant implementation of AI solutions – from selecting appropriate systems to documentation and ongoing monitoring. Contact us for a no-obligation consultation.
AI for Your Business
Want to put AI to work in your business?
From AI agents to private AI: we guide you from strategy to implementation — privacy-compliant and tailored to your processes.
Book a Free Consultation30 min · Free · No obligation
More Articles
Claude Fable 5 Is Here: Anthropic's Most Powerful AI Model Is Now Available to Everyone
Mert Bacak · July 2, 2026
The Building Block Approach: How We Implement AI That Actually Sticks
Daniel Da Cruz · June 17, 2026
Claude Connects to Microsoft 365 – Emails, Teams Chats and SharePoint Directly in Your AI Assistant
Mert Bacak · May 21, 2026