This paper explores the “dark side” of digital innovation management for sustainable development, with a focus on artificial intelligence (AI). While most research has concentrated on the positive impacts of digital innovation, this study examines the paradoxical tension that arises between sustainable value creation and sustainable value destruction when managing AI innovation to achieve sustainable development. We conceptualize a model explaining the antecedents and the nature of this tension, and we discussed seven illustrative cases that exemplify the practical applicability of the model’s elements. Our findings show that conflicting sub-objectives across environmental, social, and economic domains, along with the divergent interests of stakeholders, are key antecedents of the paradoxical tension. Furthermore, our model illustrates that sustainable value can be created by: (i) reducing grand challenges through automation in defining problems’ root cause, and (ii) mitigating grand challenges through augmentation of firms’ capabilities. However, we argue that sustainable value can be also destroyed when failing to address grand challenges or introducing new grand challenges. According to our study, this is due to predictable or unpredictable issues that arise during the design, development, or deployment of AI innovation. Finally, the discussion of our model further explores, through five theorical propositions, the specific elements of AI innovation that enable and amplify this paradoxical tension.