2026. Explainability-Guided Black-Box Adversarial Auditing for Detecting Hidden Data Poisoning in Multimodal Foundation Models. Journal of Data Intelligence and AI Systems. 1, 3 (Sep. 2026).