Exploring AI Governance: Part 1 of Navigating AI Series

AI governance, like all forms of good governance, is not a linear process that can be completed once and then ignored. It is cyclical in nature—one that needs constant evaluation and revision. This fact gains even more importance considering the growing adoption of AI across various sectors owing to its potential savings and opportunities for businesses. 
 
Despite the clear advantages offered by AI, a striking revelation from the December 2023 IAPP-EY Professionalizing Organizational AI Governance Report indicates that as many as 57% of European companies do not exercise control over their use of AI. The lack of robust AI governance while integrating it into products or internal procedures exposes these organizations to significant legal, financial, and reputational risks. 
 
Even though Swiss law presently does not lay down specific regulations concerning AI, it highlights the board's responsibility to oversee any initiatives related to this technology as part of their general duties. In addition, due to extraterritorial scope laws such as EU’s Artificial Intelligence Act, which establishes comprehensive compliance obligations for providers and users alike, both national and international considerations need attention when implementing an effective model for governing artificial intelligence systems within an organization. 
 
The management board should prioritize designing an adaptive framework for effective governance, taking into account risk-based approaches associated with each unique use case scenario involving artificial intelligence applications. Such a framework should strive towards balancing innovation with ethical imperatives alongside compliance with existing regulations whilst delivering commercial value. 
 
Key elements like assigning responsibilities appropriately throughout organizational hierarchy levels are crucial when establishing a dynamic system capable enough to govern complex technologies like artificial intelligence effectively. 
 
Roles must be clearly defined along with responsibilities corresponding to decision-making processes so as to ensure alignment between company goals or ethical standards set forth by them. Reporting structures should also be established clearly so that regular communication between different teams involved in project implementation can take place smoothly without hindrance, thereby enabling better guidance or support where necessary from key stakeholders, including legal experts, IT personnel, etc., who are part of this committee. 
 
The company bylaws and policies must reflect the responsibilities related to AI governance. This not only increases accountability but also limits potential liability under Swiss law. By clearly stating the structure of the AI governance system, division of responsibilities, and allocation of duties in these documents, companies can protect themselves from legal risks. 
 
Risk assessment is a vital component for effective AI governance. It involves identifying potential risks, such as data breaches or ethical violations, and developing strategies to mitigate them effectively through measures like encryption or anonymization, along with internal guidelines, training programs, etc., that are designed in accordance with specific needs posed by each unique scenario involving use of this technology within an organization's operational framework. 
 
Internal policies based on risk assessments need to be implemented that address both legal and non-legal ethical principles crucial for ensuring good governance over artificial intelligence systems used within any organizational setup, irrespective of the size and nature of operations conducted thereunder. 
 
Lastly, it’s important to remember that implementing AI governance requires continuous monitoring and improvement. The steps involved should be re-performed regularly based on the established framework, ensuring its effectiveness in managing AI applications responsibly while mitigating associated risks efficiently, thereby enabling organizations to leverage the benefits offered by this revolutionary technology without compromising their integrity or reputation among stakeholders at large. 
 
Contributors: Christoph Lang (Partner), Sarah Drukarch (Partner), Simon Winkler (Associate), Luise Locher (Junior Associate).