Program
Earth, as seen by Voyager 1
from 6 billion km away (1990).
Wednesday, 2 September 2026
This talk will provide a personal review of 30 years of researching how human behaviour in security, and our community’s efforts to develop effective solutions to protect individuals, organisations and societies from increasing numbers of attacks. Usable security focused on functionality and aesthetics of expert-designed solutions, human-centred security put human needs and preferences at the centre of design. The STS perspective offers a different perspective of the processes involved in security — and I will argue that re-conceptualising security in terms of care, maintenance and tinkering can help organisations.
Thursday, 3 September 2026
Large language models are increasingly used to simulate human decision-making. In this talk, I examine whether personality-aligned LLM “agents”, endowed with Big-Five profiles, can reproduce human differences in news judgement — the ability to rate true news as accurate and false news as inaccurate. Using existing datasets where human participants with known personality traits evaluated political headlines, we create matching LLM agents and compare their responses with human patterns. I will present where trait effects (e.g., Agreeableness, Conscientiousness, Open-Mindedness) are echoed by LLMs, where they diverge, and how personality conditioning changes overall accuracy ratings across different prompt formats and model settings. This offers a cautious view of what current LLMs can and cannot tell us about personality-driven susceptibility to misinformation.