To overcome the limitations of visible light communication (VLC) systems while benefiting from high data rates and inherently secure transmissions, hybrid systems combining radio frequency (RF) and VLC have gained appeal for the next generation of wireless networks and their use cases, including the Internet of Things (IoT) and sensor networks. In this article, a theoretical study is carried out to investigate the potential of hybrid systems from the security perspective by proposing a joint precoding and user association (JPUA) scheme to maximize the sum secrecy rate of a multiuser indoor downlink hybrid RF/VLC system. Specifically, it is assumed that the RF access point (AP) is equipped with multiple antennas while the VLC AP is equipped with multiple light-emitting diodes (LEDs). The optimization problem is divided into two subproblems, the user association strategy and the precoding design for each of the APs. For the former, an algorithm based on coalitional game theory is proposed to associate each user (or sensor node) with one of the APs. The latter leads to a nonconvex fractional programming problem, which is reformulated using the constrained convex-concave procedure, thus efficiently converging to an optimal solution. Numerical results based on simulations show that the proposed hybrid RF/VLC scheme provides significant gains in terms of sum secrecy rate, compared with different baseline schemes, such as standalone VLC or RF schemes, in addition to allowing the connection of a greater number of users.