LOOM: a Privacy-Preserving Linguistic Observatory of Online Misinformation

Online misinformation is an ever-growing challenge that can have a negative impact on individuals, societies, and democracies. We report on LOOM, a project that aims to build and validate a browser-based tool to detect and respond to misinformation in a trustworthy and privacy-preserving manner to protect end users and build public resilience to untrustworthy content. LOOM uses natural language processing techniques to detect and flag linguistic misinformation markers for end users in real time, whilst preserving the end-to-end encryption that protects the privacy and security of their online browsing and communication activities. We applied a citizen science framework to test the tool as an intervention to build user resilience to false information content and assess their trust in the tool. Feedback from participants indicates that the tool has the potential to improve user awareness of subtle language cues associated with misinformation and to help them critically evaluate the information they encounter. Overall, our experiment indicates a demand for tools to combat misinformation, but also highlights the challenges in creating a tool that is both effective and user-friendly.