Beyond Verification: Abductive Explanations for Post-AI Assessment of Privacy Leakage