Purpose: Clients may now disclose sensitive concerns to artificial intelligence (AI) tools before contacting a social worker, a practice we define as client AI-first disclosure. This exploratory study estimates its prevalence and examines whether social workers’ confidence in responding is associated with technical familiarity, ethical competence, or organizational guidance. Method: A cross-sectional survey of 401 registered social workers in Hong Kong assessed AI competence, AI use, wariness, training, organizational boundaries, exposure to AI-first disclosure, and response confidence. Analyses included paired-samples t tests, ordinal logistic regression, diagnostics, and sensitivity checks. Results: Among respondents who answered the item, 56.6% had encountered AI-first disclosure; 35.2% felt confident responding, and 36.4% reported clear organizational guidance. Confidence was associated with organizational boundaries and ethical competence, whereas technical proficiency, use, training, and exposure were not. Discussion: Within the limits of a cross-sectional design and single-item disclosure measures, the findings suggest a readiness gap that is linked more clearly to organizational guidance and ethical capacity than to technical familiarity alone. It carries practice and policy implications of renovating the existing training framework to address the gap.