Artificial intelligence (AI) is increasingly used across various areas of public policy, such as healthcare, security, and education. The objective of this article is to examine citizens’ trust in different AI tools applied to two policy domains: elder care and education. Using survey data from citizens in Spain (n = 1,200), we analyze trust in the use of robots, automated assistance tools, and predictive AI applications across different aspects of elder care and education services —for example, the allocation of places in nursing homes or the allocation of scholarships in primary education. Unlike previous studies that focus on the effect of sociodemographic variables on trust, our argument is that vulnerability and progressive ideology shape citizens’ trust in AI. First, we argue that citizens’ vulnerability tends to decrease their trust in these tools. Second, we also argue that there are differences within progressive ideology itself: while more traditional left-wing individuals tend to trust these tools more, those with a more left-leaning progressive orientation show lower levels of trust in AI-based applications. Our analysis controls for socioeconomic and institutional variables. The results highlight the importance of considering citizens’ vulnerability and political stances when deciding whether or not to implement AI technologies in public services.