The present study investigates the effectiveness of an GenAI-based chatbot, "Reading Bot," in reducing foreign language reading anxiety (FLRA) and improving foreign language reading performance (FLRP) among Chinese secondary school students learning English as a foreign language (EFL). To do so, a mixed-methods quasi-experimental pre-test/post-test design with qualitative interviews was employed. After ensuring homogeneity between the two groups, one class was designated as the experimental group (n=42), utilizing the chatbot as a treatment, while the other class served as the comparison group (n=42), receiving traditional teacher support as a treatment. Both groups participated in five 45-minute reading practice sessions. The results indicated that the chatbot intervention significantly reduced the participants' FLRA within the experimental group when comparing pre- and post-test scores. However, no significant differences were found between the two groups after the treatment, either in FLRA or FLRP. The qualitative analysis of semi-structured interviews suggested that the chatbot provided technological, pedagogical, linguistic, and affective affordances to assist reading. Nevertheless, the analysis also revealed some potential drawbacks and challenges to an GenAI-enhanced approach to reading instruction that may have influenced FLRA and FLRP. The theoretical and pedagogical implications of these findings are discussed, along with potential directions for future research.