
Switzerland is one of the world’s leading financial centres, and its financial sector is among the country’s most heavily regulated industries. At the heart of this framework, FINMA works to protect clients and maintain financial system stability. Cyber threats, operational resilience and artificial intelligence: its requirements are evolving rapidly and directly affect data management.
Au sommaire
- What is FINMA and how does its regulatory framework work?
- Who is subject to FINMA regulation?
- What are FINMA’s main requirements?
- What are the risks associated with using AI in the financial sector?
- What are the penalties for non-compliance?
- How does DOT Anonymizer help organisations meet FINMA requirements?
- Conclusion
1. What is FINMA and how does its regulatory framework work?
FINMA, the Swiss Financial Market Supervisory Authority, is the independent authority responsible for supervising Switzerland’s financial sector. Operational since January 2009, it was formed through the merger of the Swiss Federal Banking Commission, the Federal Office of Private Insurance and the Anti-Money Laundering Control Authority. Its mandate is to protect creditors, investors and policyholders, and to ensure that financial markets function effectively.
Unlike DORA, FINMA regulation is not a single piece of legislation. Its principles are established in laws, including the Banking Act, the Financial Services Act and the Anti-Money Laundering Act, detailed in Federal Council ordinances, and further clarified in FINMA circulars, which describe its supervisory practices.
2. Who is subject to FINMA regulation?
FINMA supervises a broad range of financial institutions, known as “supervised institutions”: approximately 250 banks, 195 insurance companies, 1,600 portfolio managers and nearly 10,000 insurance intermediaries, as well as participants in the collective investment schemes sector.
Its supervision is risk-based and proportionate: the larger and more complex an institution, the more closely it is supervised. Switzerland’s largest bank therefore undergoes more than 40 on-site reviews each year, compared with one every eight to ten years for a small bank. Banks’ IT service providers are also indirectly affected, as FINMA’s requirements extend to outsourced services and cloud computing.
3. What are FINMA’s main requirements?
In 2025, nearly half of the cyberattacks reported to FINMA targeted financial institutions’ service providers and subcontractors. In response to these threats, FINMA has strengthened its expectations. Institutions are required, in particular, to:
- Manage their operational risks and resilience, in accordance with Circular 2023/1, in force since January 2024: this includes managing IT and cyber risks, ensuring business continuity and maintaining critical functions during a crisis.
- Protect their critical data: each institution must identify this data, ensure its confidentiality, integrity and availability, and immediately report any significant incident affecting it to FINMA.
- Establish controls over their use of artificial intelligence, in accordance with Guidance 08/2024: this includes maintaining an inventory of AI systems, ensuring data quality, conducting regular testing, making results explainable and arranging independent verification.
Institutions operating across borders must also comply with the EU AI Act, which is being phased in through 2028.
4. What are the risks associated with using AI in the financial sector?
Artificial intelligence is already well established in Switzerland’s financial sector. According to a FINMA survey published in April 2025, around half of the institutions surveyed use AI, and the vast majority of those use generative AI. While AI improves operational efficiency, its use also introduces new risks. FINMA highlights several areas of concern:
- Cybersecurity risks: AI applications can increase IT systems’ exposure to cyberattacks and unauthorised access.
- Data confidentiality: using personal, banking or financial information in AI models can create a risk of disclosure.
- Algorithmic errors and bias: inaccurate data or inadequately controlled models can lead to incorrect decisions.
- Dependence on external providers: the growing use of third-party AI solutions can limit control over data and operational risks.
Data protection is therefore a central concern, particularly when institutions use data from their production environments to develop, train or test AI applications.
5. What are the penalties for non-compliance?
When supervisory law is breached, FINMA initiates enforcement proceedings and has extensive powers at its disposal. It can order interim measures, revoke licences, liquidate companies operating without authorisation or confiscate unlawfully obtained profits. It can impose industry bans on executives responsible for misconduct and publish decisions naming them. In the most serious cases, it files a criminal complaint with the competent authorities.
Stronger powers under consideration
Unlike European authorities, FINMA does not yet have the power to impose fines. This could change: on 12 August 2026, the Federal Council launched a consultation on a reform that would give FINMA the power to impose fines and periodic penalty payments, and introduce a personal accountability framework for senior executives (Senior Managers Regime) at banks with at least 250 employees. These measures still require parliamentary approval.
6. How does DOT Anonymizer help organisations meet FINMA requirements?
FINMA’s security and confidentiality requirements mean that financial institutions must exercise greater care when using sensitive data. In this context, DOT Anonymizer, ARCAD Software’s data anonymisation solution, helps protect personal and confidential information, particularly in development and testing environments.
The solution enables organisations to:
- Protect sensitive data outside production environments: avoid using real personal data during testing and development to reduce the risk of disclosure.
- Preserve data consistency: maintain realistic datasets that remain usable for business purposes.
- Reduce AI-related risks: limit the exposure of personal data when using artificial intelligence by providing AI tools and models with anonymised data instead of real data.
By reducing the exposure of sensitive data, DOT Anonymizer also helps strengthen IT system resilience and supports compliance with the Federal Act on Data Protection (FADP).
Anonymisation therefore complements the governance, cybersecurity and control measures expected by FINMA. However, it is not sufficient on its own to guarantee regulatory compliance.
7. Conclusion
FINMA plays a central role in maintaining the strength and reputation of Switzerland’s financial sector. By extending its supervision to operational resilience, cyber risks and artificial intelligence, it makes critical data protection a major compliance priority.
With the expected strengthening of its powers, particularly the ability to impose fines, the cost of non-compliance could rise significantly. Swiss financial institutions and their service providers should therefore prepare for these requirements now.
TRIAL VERSION / DEMO
Request a trial version or a session in our sandbox!
or


