Large Language Models (LLMs) have rapidly become ubiquitous, demonstrating remarkable proficiency in generating text and responding to prompts. Despite their potential, concerns about privacy and trust persist, yet methods for engaging end-users in designing privacy-conscious AI remain limited. This paper presents a replicable co-design methodology for engaging end-users in privacy-conscious LLM design. Through workshops with 36 participants across three speculative scenarios—mental health applications, travel assistants, and workplace assistants—we demonstrate how combining speculative scenarios with established frameworks (Grice’s Maxims, Schaub’s privacy design space) and trained facilitators enables meaningful participation from users without technical expertise. The co-designed dialogues reveal that users desire dynamic, context-sensitive privacy communication that leverages LLMs’ conversational capabilities. We provide practitioners with a replicable methodology comprising: (1) scenario creation methods, (2) framework scaffolding approaches, and (3) facilitator training guidance, alongside design implications for privacy-conscious conversational AI.

Since Siri’s release in 2011 there have been a growing number of AI-driven domestic voice assistants that are increasingly being integrated into devices such as smartphones and TVs. But as their presence has expanded, a range of ethical concerns have been identified around the use of voice assistants, such as the privacy implications of having devices that are always listening and the ways that these devices are integrated into the existing social order of the home. This has created a burgeoning area of research across a range of fields including computer science, social science, and psychology. This paper takes stock of the foundations and frontiers of this work through a systematic literature review of 117 papers on ethical concerns with voice assistants. In addition to analysis of nine specific areas of concern, the review measures the distribution of methods and participant demographics across the literature. We show how some concerns, such as privacy, are operationalized to a much greater extent than others like accessibility, and how study participants are overwhelmingly drawn from a small handful of Western nations. In so doing we hope to provide an outline of the rich tapestry of work around these concerns and highlight areas where current research efforts are lacking.

To improve user experience, Alexa now allows users to consent to data sharing via voice rather than directing them to the companion smartphone app. While verbal consent mechanisms for voice assistants (VAs) can increase usability, they can also undermine principles core to informed consent. We conducted a Delphi study with experts from academia, industry, and the public sector on requirements for verbal consent in VAs. Candidate requirements were drawn from the literature, regulations, and research ethics guidelines that participants rated based on their relevance to the consent process, actionability by platforms, and usability by end-users, discussing their reasoning as the study progressed. We highlight key areas of (dis)agreement between experts, deriving recommendations for regulators, skill developers, and VA platforms towards crafting meaningful verbal consent mechanisms. Key themes include approaching permissions according to the user’s ability to opt-out, minimising consent decisions, and ensuring platforms follow established consent principles.

This paper reports on empirical work conducted to study perceptions of unfair treatment caused by automated computational systems. While the pervasiveness of algorithmic bias has been widely acknowledged, and perceptions of fairness are commonly studied in Human Computer Interaction, there is a lack of research on how unfair treatment by automated computational systems is experienced by users from disadvantaged and marginalised backgrounds. There is a need for more diversification in terms of the investigated users, domains, and tasks, and regarding the strategies that users employ to reduce harm. To unpack these issues, we ran a prescreened survey of 663 participants, oversampling those with at-risk characteristics. We collected occurrences and types of conflicts regarding unfair and discriminatory treatment and systems, as well as the actions taken towards resolving these situations. Drawing on intersectional research, we combine qualitative and quantitative approaches in order to highlight the nuances around power and privilege in the perceptions of automated computational systems. Among our participants, we discuss experiences of computational essentialism, attribute-based exclusion, and expected harm. We derive suggestions to address these perceptions of unfairness as they occur.

This paper discusses how born-digital cultural material can be opened up for research. We focus in particular on the grey area between private mobile phone data and its publication and use for research and beyond. We report on the results of the `Empowering Data Citizens' (EDC) project, which is a collaboration between King's College London and the Open Data Institute. The work builds on the project Our Data Ourselves (http://big-social-data.net/), which studies the content we generate on our mobile devices, what we call big social data (BSD), and explores the possibilities of its ethical storage.

In this short paper we discuss our work on co-research devices with a young coder community, which help investigate big social data collected by mobile phones. The development was accompanied by focus groups and interviews on privacy attitudes and aims to explore how youth cultures are tracked in mobile phone data.