查看文章

Rising utilization of novel artificial intelligence and other data-driven applications sets a demand for privacypreserving large-scale data management. In the current, cloud-centric model, trust is placed on third parties that collect, aggregate, link and analyse personally identifiable information (PII) with artificial intelligence (AI) and machine learning (ML) applications, gaining unprecedented insight into individuals on the way. In the future, the amount of gathered PII will only increase, with sensors and datagathering applications becoming ever more ubiquitous. It has lately become obvious that placing trust on service providers has been unfounded. Personal information has already been used by private companies to improve their business without the consent of the individuals.

Privacy by Design (PbD) is a novel approach that ensures privacy protections are, by default, built into systems right from the start [6]. PbD can be considered a good business practice, enhancing user trust and improving user experience. Although the growth of AI and large data sets pose a great risk to privacy, the introduction of PbD will proactively embed privacy into the design of AI systems. The General Data Protection Regulation (GDPR) initiated by EU is a recent example that integrates privacy into systems using the PbD concepts. Similar approaches in the global level may advocate AI systems to prevent harm from arising and avoid data breaches.