Multi-party Updatable Delegated Private Set Intersection

Prepared by: Aydin Abadi, Changyu Dong, Steven Murdoch and Sotirios Terzis

Abstract: With the growth of cloud computing, the need arises for Private Set Intersection protocols (PSI) that can let parties outsource the storage of their private sets and securely delegate PSI computation to a cloud server. The existing delegated PSIs have two major limitations; namely, they cannot support (1) efficient updates on outsourced sets and (2) efficient PSI among multiple clients. This paper presents “Feather”, the first lightweight delegated PSI that addresses both limitations simultaneously. It lets clients independently prepare and upload their private sets to the cloud once, then delegate the computation an unlimited number of times. We implemented Feather and compared its costs with the state of the art delegated PSIs. The evaluation shows that Feather is more efficient computationally, in both update and PSI computation phases.

A copy of this paper is available for download here.

Nothing to Be Happy About: Consumer Emotions and AI

Prepared by: Mateja Durovic and Jonathan Watson

Abstract: Advancements in artificial intelligence and Big Data allow for a range of goods and services to determine and respond to a consumer’s emotional state of mind. Considerable potential surrounds the technological ability to detect and respond to an individual’s emotions, yet such technology is also controversial and raises questions surrounding the legal protection of emotions. Despite their highly sensitive and private nature, this article highlights the inadequate protection of emotions in aspects of data protection and consumer protection law, arguing that the contribution by recent proposal for an Artificial Intelligence Act is not only unsuitable to overcome such deficits but does little to support the assertion that emotions are highly sensitive.

Keywords: AI; consumer law; new technologies; regulation; emotions; EU Law

A copy of this paper is available for download here.