Public perceptions of legal chatbots and regulatory asymmetry in online discourse

This study investigates how legal chatbots are framed, contested, and evaluated in online public discourse, with a focus on trust, ethics, and regulatory expectations. Analyzing 2514 texts published on The Guardian and X (Twitter) between 2022 and 2024, the study integrates BERTopic-based topic modeling, VADER sentiment analysis, and qualitative thematic coding to capture both large-scale patterns and contextual nuance. The findings show that public perceptions of legal chatbots are not uniform but strongly shaped by application domain. Chatbots used in child protection and technology-driven services are predominantly framed as supportive and empowering tools, generating relatively higher levels of expressed trust. By contrast, deployments in government and social justice contexts are associated with skepticism, reflecting concerns about surveillance, misinformation, and the reinforcement of institutional bias. The study contributes to Algorithmic Justice scholarship by demonstrating that fairness perceptions emerge from institutional power relations rather than technical features alone, advances Digital Trust Theory by foregrounding affective dimensions of trust alongside cognitive evaluations, and introduces regulatory asymmetry as a diagnostic lens for identifying systematic gaps between ethical concern and regulatory oversight as expressed in public discourse. Together, these findings highlight the limits of one-size-fits-all AI governance and point to the necessity of context-sensitive regulatory and design approaches for AI-powered legal technologies.