As one of the most important cloud computing services, cloud storage provides storage resources for resource-constrained users, which reduces their local overhead and computing cost. As an extension of cloud computing, fog computing introduces a fog layer between the cloud and users to deploy computing, storage, and other types of equipment, allowing users to operate outsourced data conveniently. Although cloud storage brings many conveniences to users, assured data deletion is still one of the crucial security challenges. In addition, since cloud servers usually store data in a decentralized manner, it is difficult for users to centrally and securely operate data blocks. This paper proposes an efficient and secure cloud data deletion scheme (SDUS-AD) that supports dynamic data updates in multi-copy scenarios. In this scheme, a new dynamic structure is designed, which improves the traditional Merkle hash tree, thereby realizing the dynamic update of outsourced data efficiently and safely. A cloud-fog-user layer structure is used to meet the needs of resource-constrained users (such as mobile users) to update data, and a secure and trusted fog cluster is constructed using the trusted cloud platform management model based on TPM alliance to ensure the confidentiality of data privacy. Security analysis shows that SDUS-AD meets real-world security requirements. Detailed performance analysis and simulation experiments show that SDUS-AD is efficient, safe, and feasible.