Abstract:
This article argues that artificial intelligence (AI) auditing is necessary to ensure companies that adopt AI technologies comply with common social values, and that AI audit services should be guided by principles that are based on those values. However, while various bodies (such as government agencies and industry actors) have made suggestions regarding AI principles, there is currently no consensus on these principles and therefore rationalisation is imperative. Furthermore, AI and data are inexorably connected to each other, but there is currently no clear view on the relationship of data privacy principles to artificial intelligence principles.
Drawing on the literature on privacy audit principles, this paper suggests a set of fundamental principles for auditing of AI that may be applicable to organisations operating domestically or across national borders. Legislation in particular countries may assist to implement these principles but even if legislation is not adopted, auditors may apply the suggested principles which have international relevance across different jurisdictions.