5G and beyond systems call for improved service qualities. To this end, advanced communication technologies such as massive multiple-input multiple-output (MIMO) and mm-wave beamforming have been developed in recent years to use radio resources more effectively. However, efficient beamforming potentially also increases the electromagnetic field exposure (EMFE) for certain users. Thus, EMFE is currently gaining increased attention from both regulatory authorities and the public and should be part of the design criteria for evolved 5G and 6G mobile communications. Reconfigurable Intelligent Surface (RIS) has emerged as a promising future technical candidate as it can program the radio environment and obtain the desired propagation paths, which potentially could also help to control EMFE. To this end, recent research studies have addressed the beamforming design at both the base station (BS) and the RIS with limited beam power requirements as EMF constraints. In this work, treating the EMF requirement as an opportunity instead of just a limitation, we consider different cases where the non-intended user (NIU), either in between the BS and the intended user (IU) or alongside the IU, could help the network design. Specifically, we investigate the potential of EMF-constrained RIS-assisted networks with power allocation between the direct BS-IU and the in-direct BS-RIS-IU paths and efficient beamforming methods. As we presented in the analytical and numerical results, proper power allocation and beam optimization can boost the performance of the system remarkably with limited complexity and channel information.