Adaptive Approach to the Development of Standards and Best Practices for the use of Artificial Intelligence within the Swiss & Liechtenstein Financial Centre

^{Publication date: October 7, 2025}

The use cases for AI in financial services (in this context, we are referring to services provided by the financial industry, mainly banks) are manifold and the new technology is rapidly evolving and penetrating in a wide variety of forms and areas of application. Although the impression may often arise that AI will solve ‘all’ problems it is not as trivial as is often insinuated. It has already become apparent that only development and deployment of AI technology in a responsible and acceptable manner has a chance for long-term success. Ultimately, the aim is to minimise any resulting reputational risks for ‘Swiss Banking’ and thus to distinguish ourselves as a brand in a positive sense. It is therefore evident that the responsible and informed use of AI can bring great benefits and be a real business case.

Against this backdrop, law makers and regulators have already started to till this field. The regulatory community has so far pursued very diverging approaches to AI regulation, spanning from risk-based approaches to principle-based approaches and various oversight models. In some jurisdictions, regulatory intent is driven by concerns of end-users of financial services with AI embedded.

The financial industry has a long-standing history of defining best practices and related standards for the development of standards and implementation of financial services. These provisions constitute a core foundation for services that are developed, implemented, maintained and operated in an efficient and effective manner. The technical committee 68 for financial services standards (ISO/TC 68 ‘Financial Services’) at the International Organization for Standardization (ISO) is the ultimate guardian for the related financial standards and is the most senior standardisation organisation recognised by the financial industry globally.

Together, regulatory provisions as well as best practices and standards define a framework of provisions in which financial services operate. Whilst regulatory provisions tend to be of a binding nature, best practices and standards are available as non-binding provisions that the financial industry may consider for implementation in various use case scenarios.

The Swiss Association for Swift and Financial Standards (SASFS) is the representative organisation for the Financial Centre of Switzerland and Liechtenstein. The SASFS, in turn, is a registered member of ISO via the Swiss Bankers Association (SBA) and represents the Swiss/Liechtenstein Financial Centre in standardisation committees, such as the ISO/TC 68 ‘Financial Services’ mentioned above. The financial industry can span a wide spectrum of services. These include financial services a bank may offer in an intermediary role to other banks, financial services offered to end-customers (both corporate and private/retail customers), financial services offered by core banking solution providers or FinTechs to financial institutions, amongst others.

In line with its mandate, the SASFS has, early on, taken ownership for the forthcoming challenges which the introduction of AI-based techniques financial services are expected to entail. To this end, the SASFS, already in April 2025, established a dedicated SASFS Transversal Commission for Artificial Intelligence (SASFS TK AI).
 

The regulatory landscape for the application of artificial intelligence and in particular for AI applied to financial services is still very patchy.

The AI Act, proposed by the European Commission in April 2021 as part of the EU’s digital strategy, establishes a risk-based classification system for AI. This framework determines compliance requirements based on risk levels, aiming to ensure the safety of AI systems while providing legal certainty.

Switzerland tends to align its AI regulatory efforts closely with those of the EU but applies a principle-based approach, which emphasizes flexibility, proportionality, and business-friendliness.

Various combinations of the regulatory approaches mentioned above can be witnessed in further financial centres.
 

In each of the financial centres the formally binding legal and regulatory provisions can be supplemented by Best Practice recommendations and accompanying standards.

The interplay between regulatory provisions, Best Practices and standards can vary from tightly-coupled set-ups to informal arrangements or even non-coordinated arrangements. Figure 1 illustrates the variety of frameworks based on regulation, standards and Best Practices.
 

As a case in point for a more tightly-coupled set-up, the European Union could be considered in view of its Regulation (EU) No 1025/2012 on European standardisation which includes the possibility for the European Commission to request the drafting, by one of the European standardisation organisations (e.g. ESOs), of European standards in support of EU policies and legislation, referred to as harmonised European standards.

Harmonised European standards provide the technical specifications needed for a product to respect EU law. They complement and specify the requirements of the respective EU harmonised legislation.

Compliance with a harmonised standard grants products the presumption of conformity with the relevant legislation.
 

The Swiss/Liechtenstein financial Centre is exposed to a multitude of regulatory provisions, in particular to those foreign markets with existing substantial business interactions.

This puts a particular focus on third-party services available to financial institutions in Switzerland and Liechtenstein. Increasingly, such services include modules with AI embedded. Traditionally, financial services solutions were offered by service providers from within the financial industry, commonly referred to as financial services providers. Contrary to this, financial services with embedded AI functions can additionally be offered by providers of generic services with AI embedded, typically offered by the IT/tech industry.

When such services with AI embedded are deployed to provide financial services to a bank’s own customers, it is required that all provisions, including AI-related provisions, of applicable jurisdictions are fulfilled and verifiable.

This encompasses jurisdictions where services are offered and/or jurisdictions from where services are sourced from or to banks.

Figure 2 illustrates the interdependencies of services sourced from various groups of service providers.
 

In this endeavour, the complexities are characterised by different types of service, multiple layers of service provision, multiple sourcing jurisdictions, encompassing service providers from both the tech- and the financial services industries as well as various jurisdictions where service users are based, independently of whether a service user is another financial institution, a corporate customer or a private/retail client.

The composition of this landscape has a decisive influence on the fields of activity of TK AI despite its focus in line with its mandate for the financial industry in Switzerland and Liechtenstein.
 

Where AI compliance requirements exist which demonstrate a need for recommendations of how to implement such provisions, TK AI is in a position and has - with help of over forty experts from a diverse range of banks - started work to develop corresponding Best Practices.

In general, there is demand in the financial markets for Best Practices that define a common baseline for service providers and service consumers for such matters as conformity, quality and transparency aspects. In markets where such shared standards are not defined by legal or regulatory provisions, standards and Best Practices can equally respond to this need.

In analysing the AI services landscape – including cross-sectorial themes -, TK AI is empowered to work on developing standards and Best Practices to address the above need for national Best Practices as well as on international standardization bodies level such as e.g. ISO/TC 68/AG 6.

These type of standards and Best Practices will be particularly useful in cases where service providers and service users stem from different communities and industries that do not share the same regulatory basis, supervision and oversight arrangements. As this would be the case for the IT industry providing services to the financial industry. Standards and Best Practices create a lingua franca that can help clarify requirements in relation to AI-embedded services and, thus, support exchanges and interactions between consumers and providers of AI-embedded services.

Furthermore, high quality standards and Best Practices can help restrict regulatory provisions to the most fundamental guidelines, in particular in cases where a regulator acknowledges standards and Best Practices as a form of self-regulation.

For the Swiss/Liechtenstein market with its cross-border business, it is a key requirement to be able to implement and maintain a flexible and adaptive approach for the development of AI standards and Best Practices in order to match the spectrum of regulatory systems in the foreign financial markets with relevance to Switzerland and Liechtenstein. The TK AI stands ready to accept these challenges ahead and is well aware that this endeavour is more than a one-off task but instead will require an on-going effort.
 

As is the case for regulatory activity, the world of standards and best practices is characterised by established organisations, each with their balanced, proven and commonly accepted governance.

From a high level of abstraction, the national, European and global spheres can be distinguished, each with their own competent standardisation organisation, that is, the Swiss Association for Standardization (SNV), the European Committee for Standardization (CEN) and the International Organization for Standardization (ISO), respectively.

In the case of the Swiss/Liechtenstein domestic market, the SNV is afforded the role as Swiss standardisation organisation through Ordinance (946.511) on the notification of technical Specifications and Standards as well as on the tasks of the Swiss Association for Standardization (DE: Verordnung (946.511) über die Notifikation technischer Vorschriften und Normen sowie die Aufgaben der Schweizerischen Normenvereinigung).

Although it is established as a Swiss/Liechtenstein standardisation committee, the TK AI is active, beyond the Swiss/Liechtenstein domestic level, on all levels including on a European (CEN/CENELEC) and global (ISO) level.

For the Swiss/Liechtenstein financial centre, the TK AI aims to define best practices for the implementation of AI-embedded financial services that complement possible regulatory provisions for AI. To this end, it can make use of any standards and best practices that already exist or are under development at a European or global level.

Through its membership with SNV, the TK AI is qualified to participate, through delegation of Swiss/Liechtenstein experts, in any initiative at a European level at CEN and a global level at ISO.

Such involvement applies both to horizontal AI standardisation initiatives, which are industry overarching and consequently apply to many industries beyond financial services, as well as vertical financial services - specific AI standardisation initiatives.

On a European level, standardisation work for financial services (vertically) and AI standardisation work in sector-overarching industries (horizontally) are effected through CEN/TC 475 (Finance) and CEN/CLC/JTC 21 (Artificial Intelligence), respectively.

On a global level, standardisation work for financial services (vertically) and AI standardisation work in sector-overarching industries (horizontally) are the responsibility of ISO/TC 68 (Financial Services) and of ISO/IEC JTC 1/SC 42 (Artificial Intelligence), respectively.

As a complementary area focusing specifically on AI standardisation in the financial industry sphere (vertically), the two ISO/IEC committees have jointly established a Joint Working Group ISO/IEC JTC1/SC 42 - ISO/TC 68 (Financial Services): Artificial intelligence (ISO/IEC JTC 1/SC 42/JWG 7).

Figure 3 illustrates the standardisation bodies on a national, European and global level where AI standardisation occurs.
 

On all of the three “standards setting” levels, the TK AI is empowered to represent the interests of the Swiss/Liechtenstein financial centre. The Swiss/Liechtenstein banking community can form its opinion via TK AI and make its voice heard in those committees.

In this vein, the TK AI can initiate proposals to develop common standards and best practices and respond to standardisation initiatives and proposals of other stakeholders and financial centres. At the same time, it can participate in the assessment of standards and best practices and related development efforts sponsored and proposed by other financial centres, in particular, if those standards and Best Practices are of relevance to the Swiss/Liechtenstein financial market.

So far, the financial industry is still underrepresented in horizontal industry committees which tend to be dominated by the large providers of generic AI solutions and services. This makes it even more relevant to contribute demands and needs of the Swiss/Liechtenstein financial industry in this field. Otherwise, the Swiss/Liechtenstein financial market risks merely to be driven by others who might not share the same goals and values regarding minimum standards and requirements of the financial industry.

For the effort to define own standards and Best Practices, the TK AI can draw upon the vast range of collateral items, supporting material and tools which are readily available from the established standards organisations, primarily SNV, CEN and ISO. Amongst others, development processes for various document types exist, amongst others, International Standards, technical specifications (TS), publicly available standards (PAS) and technical reports. These supporting items can be freely employed for each standardisation initiative as applicable, from the classical standards or Best Practices to the more complex and comprehensive quality management system or conformity assessment solutions.

Torsten Schmidt (Chair TK AI) remarks: «The SNV provides, to the SASFS Transversal Commission for Artificial Intelligence (TK AI), a proven standardisation framework which can readily be employed for the development of Best Practice recommendations and standards for the use of AI in financial services both for our Swiss/Liechtenstein home market as well as for any relevant initiatives on regional or global scale.»

Tiphanie Bent (IT Governance Officer, SIX Group Services AG, Deputy Chair TK AI) adds: «The way SNV is connected to the relevant standardisation organisations in Europe and globally, provides us a direct access to the standard champions both in the finance sector and the IT industry. Through its involvement in TK AI and the SASFS in general, SIX fully stands behind the approach pursued by TK AI.»

Torsten Schmidt summarises: «The current set-up of TK AI, the active support received from the SASFS executive board members and their institutions as well as the contributions of the TK AI members are extremely encouraging and promise to contribute to an effective & efficient framework for deployment of secure and safe AI services for the Swiss/Liechtenstein financial centre.»

Denise Tischhauser (SASFS president) concludes: «Despite the immediate need for and focus on defining Best Practice and standards for Switzerland and Liechtenstein, the TK AI aspires long-term to generate proper and helpful guidance to the benefit of the financial institutions in Switzerland and Liechtenstein, as all members of the community are confronted with very similar challenges regarding the use and implementation of AI.»
 

In order to keep the Swiss/Liechtenstein financial standards community informed on the latest trends and to provide regular updates on current standardisation initiatives and developments, the SASFS publishes articles on relevant developments (https://www.sasfs.ch/fachartikel) and organises regular briefings, so-called ‘SASFS Quarterly Insights’ Webinars. Participation is free for all SASFS member institutions following prior registration through the SASFS website: https://www.sasfs.ch/events.

In case of general enquiries regarding the activities and the role of the SASFS, the SASFS office (SASFS Geschäftsstelle) is available as a first point of contact: https://www.sasfs.ch/kontakt.