Abstract

With the advent of GenAI technology within the Internet of Things ecosystem, research administrators now have the unprecedented ability to interact with chatbots for work-related inquiries across a wide range of personal devices, including phones, tablets, smartwatches, and laptops. This ubiquitous access, while promising significant benefits, introduces a spectrum of information security risks that have yet to be thoroughly examined. My presentation delves into three potential security vulnerabilities of GenAI chatbots (prompt injection attacks, data breach, hallucination), illustrated through fictional yet pertinent scenarios. It also proposes three policy recommendations (targeted awareness raising, regular audit, and access control) and raises four questions for future research and discussions. This presentation aims to engage the audience, encouraging reflection and inquiry.