Financial Services regulators around the globe currently work on new principles and standards for the supervision of data science, machine learning and artificial intelligence (BDAI). Although some regulatory standards have already been published, most of the work is in an emerging status, being in discussion between a variety of stakeholders (e.g. regulation experts, actuaries and quants, data scientists, IT experts with scientific, technology as well as industry backgrounds). In general, during these discussions a variety of ideas and approaches have emerged. In addition there are references to local insurance regulation, but also to IT regulation, data privacy and customer protection legislation. 

Nevertheless in all these discussions commonalities have emerged, and we would like to sketch these key ingredients of future BDAI regulation to the audience. To this end, we give an overview over relevant regulatory drafts and derive common underlying reasoning and guardrails, together with most prominent challenges. In particular we look into the areas of 

• risk and impact assessment for BDAI use cases in insurance, 
• potential governance measures, 
• data and performance monitoring, 
• algorithmic validation requirements and 
• cross-references to existing regulations like Solvency II. 

In doing so we also highlight initial results of a recently established DAV Working Party on BDAI Validation. We close the presentation by summarizing anticipated key elements of a potential future regulatory framework.