With the rapid adoption of artificial intelligence (AI) in consumer products, Product Managers (PMs) face an unprecedented responsibility: AI security. This article explores the critical role of PMs in identifying and mitigating two primary risks in AI systems: data leaks (such as potential exposure of sensitive training data through crafted prompts) and model manipulation (such as adversarial attacks that cause unintended system behaviors). We present a pragmatic, PM-centric framework for managing AI security risk that can be woven into existing product development workflows. First, PMs should facilitate threat modeling as part of the discovery process to identify potential misuse cases and inform the risk management strategy. Second, PMs can define security-oriented user stories and architectural guardrails during the design phase. Third, PMs should coordinate with security teams to perform red-teaming exercises before launch. Continuous prevention requires PMs to establish data governance as a top priority and promote consistent robustness testing practices. Success in this endeavor requires the PM to be the connective hub in the organization—translating technical risk to business risk and collaborating closely with cross-functional teams including Security, Legal, and Engineering to implement an effective security strategy. By building these elements into the fabric of how they work, PMs can position themselves as the first line of defense in upholding user trust and product integrity.