As recommendation systems in online retail increasingly embed paid placements into personalized product discovery, there is a risk that personalization becomes manipulation. This conceptual paper looks at how AI-powered ad-transparency tools can reduce manipulative personalization in e-commerce, without destroying the value of personalization. The paper integrates persuasion knowledge model, reactance theory, trust theory and stimulus-organism-response paradigm with the literature on dark patterns, explainable AI and recent regulation such as the EU Digital Services Act, the EU AI Act and India’s Guidelines for Prevention and Regulation of Dark Patterns, 2023. It defines manipulative personalization through three criteria (covertness, vulnerability exploitation and deliberation bypassing) and proposes a four-dimension typology of AI ad-transparency tools (disclosure, explanatory, control and audit transparency). It constructs a framework with three propositions: AI ad-transparency tools activate consumers’ persuasion knowledge (P1); when recognized, manipulative tactics increase perceived manipulation, stronger when they are personalized than generic (P2); and control and explanatory transparency increase perceived autonomy and algorithmic trust (P3). The paper repositions transparency from a disclosure label to a multi-layered AI capability that both exposes manipulation and restores consumer agencies, providing a transparency-by-design approach for retailers, substantive criteria for regulators, and testable propositions for future empirical research.